Traces

Every run: the agent's full trajectory, its submissions over the 20-hour budget, and the final score.

Kolmogorov Audio Compression

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.0779$45.1977.6M262k17h 49m33416
2Claude Fable 5.10.0754$52.3588.6M331k12h 35m33910
3Claude Fable 5.10.0711$71.87125.0M385k17h 43m41418
4Claude Fable 5.10.0695$54.4693.0M327k16h 46m34711
5Claude Fable 5.10.0678$47.9881.1M304k16h 17m33310
6GPT-5.60.0416$183.22165.9M262k15h 47m47112
7GPT-5.60.0408$43.3247.2M136k19h 34m24910
8GLM-5.30.0369$28.7191.1M330k19h 38m3432
9GLM-5.30.0349$48.94118.7M352k19h 6m3842
10Kimi K30.0313$38.45102.2M362k20h 1m39710
11GLM-5.30.0298$35.1891.1M311k19h 41m36613
12DeepSeek V4 Flash Vision Exp0.0268$2.15112.2M397k7h 43m3893
13Kimi K30.0266$26.1560.2M321k19h 29m30313
14GLM-5.30.0264$29.0175.1M335k17h 41m3116
15Kimi K30.0251$50.26120.4M460k19h 45m4107
16Kimi K30.0250$34.5393.2M354k19h 44m3958
17GLM-5.30.0243$29.3996.9M323k19h 43m3561
18Gemini 3.7 Flash0.0240$13.44120.0M296k7h 42m4657
19Gemini 3.7 Flash0.0232$21.26112.5M588k9h 45m34020
20DeepSeek V4 Flash Vision Exp0.0228$5.13304.7M583k13h 30m7127
21GPT-5.60.0224$49.4277.2M169k19h 53m42211
22Grok 4.60.0221$127.10131.5M617k19h 58m61730
23Grok 4.60.0192$168.00174.0M498k19h 58m76868
24Gemini 3.7 Flash0.0189$26.8583.8M686k9h 43m26716
25Grok 4.60.0171$114.81118.3M558k19h 40m59947
26Qwen3.8-Max0.0171$21.5569.7M606k19h 26m44615
27GPT-5.60.0163$47.5455.3M185k19h 43m29110
28GPT-5.60.0159$41.6061.4M164k18h 50m36811
29Qwen3.8-Max0.0155$18.8060.7M524k19h 36m4004
30Muse Spark 1.20.0126$49.13229.2M341k8h 52m5756
31Grok 4.60.0126$162.44156.4M577k19h 37m63461
32Kimi K30.0120$62.95172.7M576k20h 0m47315
33Qwen3.8-Max0.0118$16.9953.1M556k19h 0m4152
34Gemini 3.7 Flash0.0110$6.6319.2M229k5h 4m1665
35Grok 4.60.0106$206.03201.8M768k19h 59m84563
36Muse Spark 1.20.0104$27.73109.9M451k7h 13m35810
37Qwen3.8-Max0.0098$20.1554.4M994k19h 8m3216
38Muse Spark 1.20.0097$69.69161.5M457k8h 41m4415
39Gemini 3.7 Flash0.0097$3.4713.3M97k5h 22m1534
40DeepSeek V4 Flash Vision Exp0.0094$6.17373.9M627k11h 53m7989
41Qwen3.8-Max0.0089$23.5477.9M588k18h 20m4032
42Muse Spark 1.20.0087$13.2553.2M190k6h 44m2953
43DeepSeek V4 Flash Vision Exp0.0081$3.17180.3M434k7h 52m5542
44Inkling0.0073$7.838.2M21k1h 33m1412
45Muse Spark 1.20.0065$21.1865.3M269k5h 22m3133
46Inkling0.0061$5.435.5M41k1h 2m1472
47Inkling0.0061$13.6714.2M40k2h 24m2252
48DeepSeek V4 Flash Vision Exp0.0056$3.93222.7M553k14h 38m5307
49Inkling0.0039$4.624.8M24k1h 6m1011
50Inkling0.0030$5.705.9M31k2h 11m1451

Astronomy Toolkit

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.8953$64.06109.1M391k4h 49m30517
2Claude Fable 5.10.8935$57.2297.5M350k6h 25m29824
3Claude Fable 5.10.8000$36.7659.7M285k3h 53m22011
4Claude Fable 5.10.7999$32.0151.0M267k3h 57m20212
5Claude Fable 5.10.7955$42.1068.8M317k4h 11m23212
6Gemini 3.7 Flash0.7294$8.3949.1M408k2h 12m2181
7GLM-5.30.6985$72.52246.9M707k19h 16m5996
8Kimi K30.6250$60.98152.4M478k16h 31m41412
9Grok 4.60.5299$15.5715.6M323k1h 52m1064
10GPT-5.60.4849$43.3852.6M186k3h 16m2635
11GLM-5.30.4028$57.26195.8M552k12h 7m4268
12GLM-5.30.4017$72.89226.5M514k17h 29m4766
13Gemini 3.7 Flash0.4014$3.0923.0M225k38m1533
14Grok 4.60.3993$48.1146.2M395k3h 31m2086
15Kimi K30.3983$90.41238.0M505k20h 1m58720
16Kimi K30.3959$20.0846.3M279k6h 2m2055
17GLM-5.30.3930$77.97220.3M559k15h 22m4812
18Kimi K30.3804$70.49167.8M452k19h 35m4688
19Kimi K30.2995$53.30137.6M418k11h 54m4307
20GLM-5.30.2893$64.65196.5M549k19h 1m4198
21Grok 4.60.2000$34.6834.5M354k2h 42m1758
22Grok 4.60.1971$39.6635.3M400k3h 43m16511
23GPT-5.60.1823$39.2848.9M195k2h 7m2554
24Muse Spark 1.20.1013$11.0759.7M393k2h 7m2591
25Muse Spark 1.20.1000$12.1666.0M402k2h 34m2314
26Muse Spark 1.20.0563$26.3390.8M427k5h 8m2972
27Qwen3.8-Max0.0000$42.82145.6M925k17h 28m5242
28Inkling0.0000$4.594.7M24k22m1052
29DeepSeek V4 Flash Vision Exp0.0000$4.02226.3M587k18h 16m5035
30Inkling0.0000$5.986.1M33k29m1441
31DeepSeek V4 Flash Vision Exp0.0000$4.14229.6M637k20h 0m4831
32GPT-5.60.0000$29.7437.3M170k2h 5m2118
33Qwen3.8-Max0.0000$48.24157.1M1.3M19h 42m5182
34Gemini 3.7 Flash0.0000$5.4438.0M299k1h 47m1973
35Inkling0.0000$2.972.9M45k43m1092
36Muse Spark 1.20.0000$53.10223.4M907k9h 16m5353
37DeepSeek V4 Flash Vision Exp0.0000$3.11165.3M552k9h 15m4096
38Muse Spark 1.20.0000$11.0558.0M328k3h 11m2412
39Qwen3.8-Max0.0000$42.25133.5M1.3M19h 46m4122
40GPT-5.60.0000$23.2833.7M166k2h 33m2075
41DeepSeek V4 Flash Vision Exp0.0000$3.44189.2M548k10h 29m4534
42GPT-5.60.0000$47.1154.0M192k2h 34m2624
43Inkling0.0000$6.496.6M47k1h 53m1421
44Qwen3.8-Max0.0000$56.35191.3M1.2M19h 55m5891
45Grok 4.60.0000$23.0923.9M317k4h 4m1523
46Gemini 3.7 Flash0.0000$3.9124.5M212k1h 28m1552
47Gemini 3.7 Flash0.0000$3.8223.6M220k46m1622
48Inkling0.0000$6.066.2M36k1h 0m1182
49Qwen3.8-Max0.0000$28.3286.1M1.0M14h 43m3863
50DeepSeek V4 Flash Vision Exp0.0000$3.67205.7M547k18h 16m5103

Cranelift Codegen Optimization

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.3484$166.03303.2M604k16h 29m82513
2Claude Fable 5.10.3356$233.87425.4M876k16h 37m102118
3Claude Fable 5.10.2852$199.91365.0M723k15h 17m97415
4Claude Fable 5.10.2009$170.71311.4M625k17h 11m81317
5Gemini 3.7 Flash0.0802$35.86231.6M242k6h 15m57420
6Kimi K30.0603$148.59441.7M786k16h 22m102815
7Kimi K30.0478$126.89362.2M801k17h 53m9369
8DeepSeek V4 Flash Vision Exp0.0237$8.27504.6M723k9h 18m126011
9Gemini 3.7 Flash0.0189$20.57157.5M204k2h 49m43217
10GPT-5.60.0176$225.59218.6M216k6h 56m5066
11GLM-5.30.0162$93.99346.3M474k11h 28m7662
12Kimi K30.0132$145.36415.9M1.0M20h 3m10797
13GLM-5.30.0129$109.37311.0M484k17h 22m7709
14Qwen3.8-Max0.0117$41.08108.0M2.2M18h 38m4846
15Gemini 3.7 Flash0.0110$23.00164.8M188k4h 11m49519
16DeepSeek V4 Flash Vision Exp0.0107$5.01303.8M439k7h 50m7238
17GLM-5.30.0102$60.68223.7M392k7h 14m6154
18DeepSeek V4 Flash Vision Exp0.0030$5.08311.7M419k8h 3m7816
19Grok 4.60.0000$113.60103.4M918k9h 35m4266
20Inkling0.0000$8.388.8M12k55m941
21GPT-5.60.0000$297.21276.6M238k7h 17m5495
22Muse Spark 1.20.0000$33.79127.0M205k10h 23m3565
23Muse Spark 1.20.0000$18.72115.3M194k3h 23m3504
24Qwen3.8-Max0.0000$34.8881.5M2.3M20h 0m4131
25Kimi K30.0000$136.80389.4M1.1M20h 0m100619
26Gemini 3.7 Flash0.0000$30.61210.1M211k5h 15m59321
27GPT-5.60.0000$234.61222.5M240k7h 38m4615
28Grok 4.60.0000$240.25234.7M1.3M16h 47m89524
29Inkling0.0000$13.7514.4M22k1h 40m1352
30Kimi K30.0000$136.72403.0M752k19h 8m114711
31Inkling0.0000$11.7912.4M15k1h 2m1211
32Muse Spark 1.20.0000$87.58303.3M711k10h 13m5625
33GPT-5.60.0000$212.19205.4M208k5h 52m4615
34Inkling0.0000$8.248.6M21k30m1042
35DeepSeek V4 Flash Vision Exp0.0000$3.68220.8M344k8h 44m6708
36Inkling0.0000$13.8214.5M21k59m1372
37Grok 4.60.0000$138.74131.9M1.2M8h 47m5876
38Claude Fable 5.10.0000$232.11427.5M765k16h 0m102335
39Grok 4.60.0000$138.58127.2M927k7h 29m4889
40Muse Spark 1.20.0000$76.77223.4M655k10h 10m4507
41Muse Spark 1.20.0000$49.18242.6M532k8h 28m4959
42Qwen3.8-Max0.0000$43.78119.5M2.2M18h 46m4962
43Qwen3.8-Max0.0000$50.74159.1M1.7M16h 16m6244
44GLM-5.30.0000$110.08353.7M515k18h 43m76118
45Gemini 3.7 Flash0.0000$47.46341.2M378k8h 44m82822
46Grok 4.60.0000$108.30103.9M740k6h 37m4476
47DeepSeek V4 Flash Vision Exp0.0000$4.87296.6M422k6h 43m7303
48GLM-5.30.0000$69.79258.7M392k7h 49m7037
49GPT-5.60.0000$355.12344.3M254k10h 32m6273
50Qwen3.8-Max0.0000$42.54115.4M2.2M20h 0m4911

Dart Style in Haskell

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.9491$569.861069.9M1.4M17h 10m228528
2Claude Fable 5.10.9284$552.061031.1M1.5M18h 56m209146
3Claude Fable 5.10.9275$456.14848.4M1.3M18h 57m202639
4Claude Fable 5.10.8919$415.43774.8M1.2M19h 19m179647
5Claude Fable 5.10.8712$313.43577.0M1.0M18h 15m128312
6GPT-5.60.7949$998.14972.7M983k19h 18m214616
7GPT-5.60.6779$433.00423.6M391k6h 1m8444
8Gemini 3.7 Flash0.6517$199.282123.1M2.6M19h 55m491762
9GPT-5.60.6344$89.8091.5M201k1h 55m3625
10Muse Spark 1.20.6241$2.2911.0M77k1h 6m1283
11GPT-5.60.6068$48.4752.9M204k1h 35m2743
12Grok 4.60.5864$513.86512.7M2.7M20h 0m205459
13Kimi K30.5650$209.81607.6M1.3M20h 1m145419
14Kimi K30.5395$190.44541.2M1.2M19h 50m13363
15Gemini 3.7 Flash0.5376$121.241110.9M1.5M19h 27m252437
16GLM-5.30.5299$165.74599.8M1.1M19h 6m139851
17Kimi K30.4874$196.59547.1M1.3M19h 46m12418
18Qwen3.8-Max0.4828$83.97303.2M1.2M16h 38m8932
19GLM-5.30.4696$158.77595.9M645k19h 33m117712
20GLM-5.30.4653$144.84516.1M611k19h 39m106622
21GLM-5.30.4595$150.71560.0M706k19h 32m126335
22GPT-5.60.4547$71.7984.5M237k2h 25m4067
23Gemini 3.7 Flash0.4365$75.56629.3M934k19h 55m13313
24Kimi K30.4305$123.27338.1M738k19h 52m6701
25Gemini 3.7 Flash0.4242$33.92324.0M891k3h 25m7808
26DeepSeek V4 Flash Vision Exp0.4014$18.121122.6M1.6M19h 51m260223
27GLM-5.30.3788$98.82359.7M600k20h 0m7381
28Qwen3.8-Max0.3212$156.83579.3M1.7M20h 1m15061
29Gemini 3.7 Flash0.3137$25.24282.5M578k1h 28m5902
30Grok 4.60.3015$389.00386.7M2.2M19h 53m164542
31Qwen3.8-Max0.2726$83.47304.6M1.0M17h 28m8432
32Qwen3.8-Max0.2674$50.86177.6M941k18h 25m66212
33DeepSeek V4 Flash Vision Exp0.2342$7.03436.2M602k8h 13m9143
34DeepSeek V4 Flash Vision Exp0.2234$21.231304.8M2.0M15h 38m28367
35Grok 4.60.1785$547.31552.2M2.4M19h 52m217774
36DeepSeek V4 Flash Vision Exp0.1638$15.31956.5M1.3M19h 4m20233
37Muse Spark 1.20.1043$2.539.5M123k1h 21m1011
38Muse Spark 1.20.1040$2.1810.0M104k1h 4m1132
39Muse Spark 1.20.1040$3.4116.0M109k59m1102
40Qwen3.8-Max0.0853$104.77373.0M1.7M19h 28m9571
41DeepSeek V4 Flash Vision Exp0.0766$15.69953.5M1.5M16h 48m21995
42Muse Spark 1.20.0377$5.5530.5M146k1h 16m1742
43Inkling0.0000$2.953.1M10k11m872
44Inkling0.0000$20.4921.4M45k32m2411
45Grok 4.60.0000$485.05493.3M2.2M19h 52m1992109
46Grok 4.60.0000$402.09394.7M2.1M17h 23m152478
47Inkling0.0000$19.8820.7M49k1h 2m2971
48Inkling0.0000$7.808.1M20k1h 3m1401
49Inkling0.0000$4.624.8M11k1h 10m1012
50Kimi K30.0000$86.57238.4M589k20h 1m5931

FFmpeg libswscale Optimization

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.8080$283.04487.0M1.6M13h 9m107046
2Claude Fable 5.10.5509$202.06343.4M1.3M9h 13m81928
3Claude Fable 5.10.5106$148.93254.0M906k6h 11m64118
4GLM-5.30.1752$226.40818.4M2.0M19h 33m188418
5GLM-5.30.1410$148.35524.2M1.7M16h 26m13038
6GLM-5.30.1369$139.76505.8M1.2M11h 14m10979
7Claude Fable 5.10.1368$219.11385.2M1.1M10h 27m83522
8Kimi K30.1260$127.79342.6M1.2M13h 59m82713
9Kimi K30.1252$211.22606.6M1.4M19h 43m13823
10GLM-5.30.1231$175.55619.5M1.7M16h 47m15098
11Kimi K30.0799$127.97356.2M1.1M15h 4m91411
12GLM-5.30.0783$177.95625.3M1.7M17h 1m14208
13DeepSeek V4 Flash Vision Exp0.0668$8.52493.3M1.0M6h 44m10958
14Kimi K30.0639$138.03382.3M1.1M17h 32m9022
15DeepSeek V4 Flash Vision Exp0.0636$9.11528.7M1.1M6h 10m11725
16DeepSeek V4 Flash Vision Exp0.0602$13.77800.3M1.6M9h 14m183119
17DeepSeek V4 Flash Vision Exp0.0483$9.39554.1M1.0M5h 54m119716
18Qwen3.8-Max0.0373$86.44286.2M2.3M17h 43m7066
19Grok 4.60.0360$419.25404.2M2.4M19h 56m1517114
20Qwen3.8-Max0.0181$64.05193.9M2.4M18h 48m5551
21Gemini 3.7 Flash0.0125$20.83184.0M617k2h 35m5242
22Qwen3.8-Max0.0053$46.03124.6M2.3M19h 58m4241
23Gemini 3.7 Flash0.0032$19.94167.8M795k2h 4m4967
24Grok 4.60.0000$1121.421126.0M2.4M18h 11m3693434
25Muse Spark 1.20.0000$15.4290.0M327k41m2862
26Grok 4.60.0000$68.4059.6M603k3h 50m2294
27Muse Spark 1.20.0000$16.9785.6M384k1h 38m2274
28Claude Fable 5.10.0000$217.24363.8M1.5M10h 22m89327
29Qwen3.8-Max0.0000$54.69158.2M2.4M18h 58m4751
30Grok 4.60.0000$132.56129.2M778k5h 28m56227
31Inkling0.0000$11.2411.7M27k1h 20m1652
32Inkling0.0000$15.4516.1M35k22m1882
33Qwen3.8-Max0.0000$30.1266.3M2.1M19h 55m2981
34Inkling0.0000$39.6641.4M77k41m3542
35Gemini 3.7 Flash0.0000$25.27228.2M1.1M2h 14m6555
36GPT-5.60.0000$66.8275.1M179k1h 59m3188
37Inkling0.0000$15.5116.2M36k32m1901
38Muse Spark 1.20.0000$11.9064.7M340k1h 14m2232
39GPT-5.60.0000$78.4384.8M190k2h 4m3485
40Grok 4.60.0000$380.75368.6M1.8M11h 52m137067
41Muse Spark 1.20.0000$35.70203.8M978k3h 29m4688
42GPT-5.60.0000$123.31124.9M233k2h 12m4223
43Kimi K30.0000$171.14500.1M1.1M19h 41m117547
44GPT-5.60.0000$76.6786.4M186k3h 5m36910
45DeepSeek V4 Flash Vision Exp0.0000$12.34718.0M1.4M8h 32m167816
46GPT-5.60.0000$117.96120.8M233k3h 42m4058
47Inkling0.0000$11.1711.6M27k16m1671
48Gemini 3.7 Flash0.0000$16.99131.7M937k2h 45m4486
49Muse Spark 1.20.0000$27.98159.1M568k1h 40m3452
50Gemini 3.7 Flash0.0000$44.74251.9M922k8h 1m7586

Crash-Proof Flash Filesystem

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.9970$49.3477.1M443k2h 57m2086
2Claude Fable 5.10.9927$43.7270.9M340k3h 17m2294
3Claude Fable 5.10.9927$61.85100.4M479k3h 5m24310
4Claude Fable 5.10.8898$81.53137.5M526k5h 0m32610
5Grok 4.60.0610$858.44859.7M2.4M19h 46m2990501
6Kimi K30.0431$197.20550.4M1.5M20h 1m11971
7Grok 4.60.0241$445.27424.9M2.8M19h 59m1572200
8GPT-5.60.0225$66.2368.5M180k1h 47m2748
9GPT-5.60.0225$81.1979.4M199k1h 55m2917
10GPT-5.60.0224$73.4072.5M171k1h 41m2738
11Gemini 3.7 Flash0.0217$93.80813.5M2.3M19h 56m21411
12Claude Fable 5.10.0173$56.4490.9M453k4h 18m2288
13GPT-5.60.0173$159.04153.2M259k4h 31m44016
14GPT-5.60.0173$92.7287.0M212k1h 59m2687
15GLM-5.30.0173$171.41627.5M1.2M19h 32m138514
16GLM-5.30.0173$236.98874.4M1.4M19h 51m190831
17Grok 4.60.0117$262.00255.9M1.7M13h 11m103920
18Kimi K30.0112$202.76569.3M1.5M19h 50m12372
19Gemini 3.7 Flash0.0112$89.95756.9M1.7M20h 0m19071
20GLM-5.30.0111$148.21526.6M1.0M18h 49m117516
21DeepSeek V4 Flash Vision Exp0.0110$29.121761.2M2.9M19h 19m391124
22DeepSeek V4 Flash Vision Exp0.0053$22.611343.5M2.5M19h 20m286726
23GLM-5.30.0049$108.31386.0M1.3M16h 54m8515
24DeepSeek V4 Flash Vision Exp0.0049$20.651205.0M2.5M19h 56m249819
25Muse Spark 1.20.0049$36.55207.9M989k3h 16m45611
26Kimi K30.0049$158.40411.0M1.7M20h 0m10101
27DeepSeek V4 Flash Vision Exp0.0049$17.941097.8M1.6M18h 58m234814
28Gemini 3.7 Flash0.0049$62.15472.1M1.5M20h 0m11391
29DeepSeek V4 Flash Vision Exp0.0049$20.781251.0M2.1M19h 4m280415
30Kimi K30.0049$190.49508.0M1.5M19h 54m11119
31Muse Spark 1.20.0047$53.94336.5M554k4h 31m6607
32Grok 4.60.0046$267.41252.9M1.7M10h 33m927132
33GLM-5.30.0039$154.89566.6M1.0M19h 45m135018
34Muse Spark 1.20.0038$13.6058.8M346k1h 50m1893
35Muse Spark 1.20.0033$35.75222.8M379k2h 39m4897
36Grok 4.60.0031$181.59175.3M1.4M8h 20m69324
37Gemini 3.7 Flash0.0024$124.701220.5M2.9M20h 0m27951
38Qwen3.8-Max0.0013$12.1816.0M1.3M19h 33m1301
39Gemini 3.7 Flash0.0010$64.73520.0M1.4M20h 0m13561
40Kimi K30.0008$195.88529.5M1.7M20h 0m12662
41Qwen3.8-Max0.0001$13.8119.3M1.5M19h 35m1643
42Qwen3.8-Max0.0000$14.3919.8M1.5M20h 1m1601
43Inkling0.0000$14.1614.7M37k1h 30m1842
44Qwen3.8-Max0.0000$9.1810.7M1.0M19h 50m1001
45Inkling0.0000$9.309.7M22k11m1232
46Inkling0.0000$12.0812.6M23k14m1612
47Qwen3.8-Max0.0000$13.7218.7M1.5M18h 42m1463
48Muse Spark 1.20.0000$18.5097.7M207k1h 10m2694
49Inkling0.0000$16.4517.2M21k16m1732
50Inkling0.0000$13.1713.7M28k1h 27m1862

FrogsGame Post-Training

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.9740$14.4121.9M144k17h 0m1535
2Claude Fable 5.10.8920$12.7119.5M122k16h 12m1464
3GLM-5.30.8380$21.9547.4M206k19h 31m2332
4Claude Fable 5.10.7800$16.4025.5M151k15h 15m1653
5GLM-5.30.7220$12.9144.6M176k19h 17m2472
6Claude Fable 5.10.6180$15.2423.4M145k17h 43m1626
7Kimi K30.4620$12.7324.9M159k19h 34m1721
8Claude Fable 5.10.4080$13.0419.3M141k15h 46m1354
9GLM-5.30.3120$16.4057.5M195k19h 21m3142
10Kimi K30.3000$16.0038.1M240k20h 0m1871
11Grok 4.60.2460$18.5719.8M283k19h 59m1521
12Grok 4.60.2420$9.6812.9M262k19h 22m1322
13Qwen3.8-Max0.2160$11.9738.7M277k19h 34m2973
14Kimi K30.1900$13.6332.0M177k20h 0m1981
15GLM-5.30.1640$20.8241.0M222k19h 4m2143
16GLM-5.30.1580$16.5958.5M233k19h 38m2651
17Grok 4.60.1280$47.6745.1M287k18h 35m2595
18DeepSeek V4 Flash Vision Exp0.1100$1.2765.0M240k20h 4m3061
19Kimi K30.1040$11.1320.9M147k19h 47m1741
20Qwen3.8-Max0.0980$10.6932.2M339k19h 46m2721
21GPT-5.60.0900$19.4628.7M100k17h 8m2062
22Qwen3.8-Max0.0880$5.8516.5M229k19h 7m1951
23DeepSeek V4 Flash Vision Exp0.0860$1.4779.6M227k19h 17m3902
24Kimi K30.0660$8.1017.2M123k19h 18m1531
25Grok 4.60.0640$49.0243.5M397k19h 5m2214
26Gemini 3.7 Flash0.0620$5.0314.3M76k5h 27m1281
27GPT-5.60.0540$20.8732.1M115k12h 56m2413
28Gemini 3.7 Flash0.0420$30.7272.9M130k16h 34m3011
29GPT-5.60.0380$22.1134.6M103k18h 7m2232
30GPT-5.60.0360$61.3164.5M145k15h 2m2772
31Qwen3.8-Max0.0360$35.74120.9M650k19h 55m5462
32Grok 4.60.0340$48.2946.2M363k16h 43m2375
33Gemini 3.7 Flash0.0320$45.33138.6M173k19h 58m3961
34Qwen3.8-Max0.0300$11.9837.6M292k18h 44m3234
35Muse Spark 1.20.0300$18.3645.0M166k4h 11m1765
36Gemini 3.7 Flash0.0220$3.8024.6M101k2h 41m1422
37Muse Spark 1.20.0220$21.7448.9M286k5h 11m23911
38Gemini 3.7 Flash0.0200$37.0796.2M140k17h 26m3211
39DeepSeek V4 Flash Vision Exp0.0160$1.3368.9M247k19h 56m3452
40DeepSeek V4 Flash Vision Exp0.0160$1.1355.9M221k19h 37m2642
41GPT-5.60.0120$91.1851.7M86k17h 41m2284
42Muse Spark 1.20.0100$18.40105.2M433k4h 43m2515
43Inkling0.0080$2.602.6M22k28m842
44Muse Spark 1.20.0060$13.7365.6M312k5h 25m1839
45Inkling0.0000$8.378.7M31k38m1692
46DeepSeek V4 Flash Vision Exp0.0000$1.6187.5M249k20h 2m3951
47Inkling0.0000$3.543.7M16k24m972
48Inkling0.0000$1.631.6M18k24m712
49Muse Spark 1.20.0000$196.241245.9M587k4h 6m23898
50Inkling0.0000$2.522.6M17k28m1032

Stepper Music Sequencer GBA

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.9662$443.40798.0M1.8M18h 14m180227
2Claude Fable 5.10.9569$440.73799.7M1.7M15h 10m182514
3Claude Fable 5.10.9473$365.73654.8M1.6M15h 4m152120
4Claude Fable 5.10.9459$457.39831.4M1.7M14h 37m184932
5Claude Fable 5.10.9053$549.231015.6M1.7M16h 2m223418
6GPT-5.60.7701$944.54934.7M1.1M15h 1m241121
7GPT-5.60.7406$409.56413.9M542k8h 34m125115
8GPT-5.60.7219$253.31245.2M326k3h 19m6796
9GLM-5.30.6108$356.801301.6M3.0M19h 1m267119
10GPT-5.60.6048$566.59550.5M752k14h 19m151311
11Grok 4.60.5671$305.84301.9M2.2M11h 0m122745
12Kimi K30.5098$230.68643.2M1.9M20h 0m15453
13GPT-5.60.4981$490.17481.7M531k5h 24m127432
14GLM-5.30.4715$281.501008.4M2.3M16h 6m226416
15GLM-5.30.4046$276.191010.3M1.8M15h 31m207021
16Kimi K30.3706$227.60629.2M1.9M19h 47m15042
17GLM-5.30.3427$275.261018.0M1.7M17h 39m209039
18Grok 4.60.3420$868.26872.5M3.4M16h 30m3114373
19Kimi K30.3170$227.44622.4M1.9M19h 52m14391
20Grok 4.60.2607$223.96223.8M1.5M7h 20m88218
21Grok 4.60.2416$205.23195.4M1.9M7h 34m79228
22Gemini 3.7 Flash0.2415$49.29499.9M1.0M4h 48m13022
23Grok 4.60.2197$129.50120.8M2.0M7h 52m54019
24Gemini 3.7 Flash0.1895$40.63436.9M982k3h 1m11342
25Gemini 3.7 Flash0.1724$88.83943.5M2.1M6h 27m24491
26Gemini 3.7 Flash0.1698$29.14318.3M646k2h 19m8562
27GLM-5.30.1440$63.57232.5M498k3h 14m4876
28DeepSeek V4 Flash Vision Exp0.1348$19.481192.0M1.8M11h 51m26707
29Kimi K30.1336$196.38508.8M1.8M19h 51m12094
30Gemini 3.7 Flash0.1224$55.17592.3M1.3M5h 23m15075
31DeepSeek V4 Flash Vision Exp0.1138$28.551772.4M2.4M15h 1m383323
32Kimi K30.0909$184.74495.2M1.5M19h 58m11542
33DeepSeek V4 Flash Vision Exp0.0534$18.461143.5M1.6M13h 42m270016
34DeepSeek V4 Flash Vision Exp0.0419$18.931164.8M1.7M11h 44m276429
35DeepSeek V4 Flash Vision Exp0.0319$17.221033.6M1.8M11h 35m230613
36Qwen3.8-Max0.0240$114.04338.0M2.8M19h 40m8846
37Qwen3.8-Max0.0224$119.80401.3M2.9M20h 1m934
38Qwen3.8-Max0.0183$113.87373.6M3.0M19h 42m990
39Muse Spark 1.20.0175$37.47155.3M1.3M5h 29m4607
40Qwen3.8-Max0.0159$129.76393.1M2.8M19h 53m1102
41Qwen3.8-Max0.0002$139.67464.7M2.7M19h 56m1203
42Inkling0.0000$3.823.9M21k15m1032
43Inkling0.0000$6.626.9M28k22m1681
44Inkling0.0000$8.488.8M35k24m1712
45Muse Spark 1.20.0000$44.79140.1M936k4h 21m5434
46Inkling0.0000$10.6411.1M32k19m1931
47Muse Spark 1.20.0000$4.6818.3M123k58m1562
48Muse Spark 1.20.0000$53.1995.9M776k4h 50m3512
49Inkling0.0000$5.725.9M31k22m1402
50Muse Spark 1.20.0000$10.5943.3M223k2h 38m2593

† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.

Git to Zig

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.5599$663.301199.7M2.6M18h 57m267417
2Claude Fable 5.10.5146$566.081019.0M2.3M19h 11m236013
3Claude Fable 5.10.5049$432.88762.8M2.1M18h 58m173213
4Claude Fable 5.10.4886$363.29631.2M2.0M18h 58m156211
5Claude Fable 5.10.4844$390.64686.7M2.0M19h 13m161126
6GPT-5.60.3440$1238.081225.0M944k19h 8m262721
7GPT-5.60.2804$1082.721080.1M850k19h 36m231442
8GPT-5.60.2801$1175.971171.4M884k19h 37m260229
9GPT-5.60.2369$955.56945.7M722k19h 4m198213
10GPT-5.60.1874$770.31763.5M629k19h 0m16686
11Gemini 3.7 Flash0.1808$157.011517.8M2.4M19h 43m32561
12Kimi K30.1768$251.00752.4M1.2M20h 2m16191
13Kimi K30.1733$200.20602.7M965k19h 43m148313
14Grok 4.60.1693$652.70657.8M2.2M19h 59m2502122
15Grok 4.60.1618$614.44610.4M2.6M18h 0m232473
16Kimi K30.1616$254.18761.8M1.1M19h 47m16689
17Kimi K30.1574$259.84779.9M1.1M19h 53m17652
18GLM-5.30.1556$225.36842.0M964k19h 45m201217
19GLM-5.30.1531$195.54692.6M689k19h 54m14802
20Grok 4.60.1516$806.68819.7M2.2M19h 59m2981186
21Inkling0.1510$17.7718.6M32k1h 29m2551
22Gemini 3.7 Flash0.1495$79.59838.0M1.2M20h 0m18552
23Kimi K30.1473$222.64660.4M1.1M19h 44m14911
24Gemini 3.7 Flash0.1433$101.571069.8M1.5M19h 58m23721
25GLM-5.30.1390$212.77708.2M852k19h 49m164049
26GLM-5.30.1378$248.79932.5M906k19h 55m221921
27Grok 4.60.1352$482.88483.0M1.6M11h 22m176784
28DeepSeek V4 Flash Vision Exp0.1304$28.181759.3M2.2M19h 24m372315
29Grok 4.60.1269$408.61405.7M1.8M19h 55m155848
30Gemini 3.7 Flash0.1237$71.55746.3M1.3M19h 57m16904
31GLM-5.30.1219$188.86712.0M621k19h 43m131816
32DeepSeek V4 Flash Vision Exp0.1213$28.061756.1M2.2M19h 43m379311
33Qwen3.8-Max0.1212$184.24696.5M1.3M19h 33m180512
34DeepSeek V4 Flash Vision Exp0.1193$26.161630.6M2.1M19h 37m35848
35DeepSeek V4 Flash Vision Exp0.1079$21.831364.0M1.7M17h 17m29778
36Qwen3.8-Max0.1038$196.74742.9M1.4M19h 38m183126
37Qwen3.8-Max0.1023$180.44685.7M1.2M19h 14m17762
38DeepSeek V4 Flash Vision Exp0.0988$20.501272.8M1.7M19h 17m273211
39Qwen3.8-Max0.0987$178.20662.0M1.7M19h 50m17106
40Qwen3.8-Max0.0927$190.64725.2M1.3M19h 52m162035
41Muse Spark 1.20.0185$42.33250.7M911k3h 1m5785
42Inkling0.0093$17.5218.3M28k1h 29m2083
43Inkling0.0019$16.3617.1M30k17m2801
44Muse Spark 1.20.0000$2.459.3M123k59m862
45Inkling0.0000$3.153.3M13k5m862
46Muse Spark 1.20.0000$5.5326.0M135k1h 56m1754
47Muse Spark 1.20.0000$59.99358.3M658k4h 22m7791
48Gemini 3.7 Flash0.0000$76.41819.6M1.5M19h 43m17341
49Inkling0.0000$5.575.8M17k1h 55m1182
50Muse Spark 1.20.0000$2.147.7M83k1h 10m933

† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.

Granite Mamba2 Inference Optimization

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1GLM-5.30.8927$85.79297.0M636k6h 7m6347
2GLM-5.30.8719$51.39174.8M618k4h 8m3828
3DeepSeek V4 Flash Vision Exp0.8708$2.68149.8M376k16h 23m4735
4Claude Fable 5.10.8458$138.31228.0M1000k9h 1m58015
5GLM-5.30.8401$45.93158.7M655k5h 28m3436
6Claude Fable 5.10.8393$189.96320.0M1.2M8h 30m71828
7Claude Fable 5.10.8171$168.52281.0M1.2M8h 8m66716
8Muse Spark 1.20.8168$5.1625.8M238k1h 23m1693
9Qwen3.8-Max0.8086$63.86200.9M2.1M18h 31m5722
10Claude Fable 5.10.8047$204.26349.5M1.2M8h 27m78431
11Inkling0.7968$5.415.6M24k1h 10m983
12Inkling0.7910$2.642.7M12k47m612
13DeepSeek V4 Flash Vision Exp0.7876$3.53195.8M540k19h 32m5198
14Qwen3.8-Max0.7876$54.77177.9M1.6M16h 1m5205
15Kimi K30.7872$122.78337.3M986k19h 10m77136
16Claude Fable 5.10.7867$157.78258.9M1.2M7h 40m58211
17Gemini 3.7 Flash0.7839$8.9784.4M442k56m2813
18Kimi K30.7805$103.11286.2M907k14h 49m65110
19Inkling0.7670$5.685.8M36k1h 22m1012
20GPT-5.60.7649$74.7474.2M207k2h 36m2705
21DeepSeek V4 Flash Vision Exp0.7578$8.78501.2M1.2M13h 5m11573
22Kimi K30.7571$91.87248.8M620k20h 0m53150
23GLM-5.30.7528$38.94126.7M486k4h 14m3515
24Kimi K30.7477$74.48205.5M639k8h 3m45620
25DeepSeek V4 Flash Vision Exp0.7467$8.27472.9M1.1M11h 58m114023
26GLM-5.30.7402$85.05310.6M673k18h 16m6458
27Grok 4.60.7305$58.7452.8M626k4h 20m20613
28Grok 4.60.7245$46.6739.7M692k3h 57m16910
29Qwen3.8-Max0.7188$63.39207.4M1.7M17h 14m5579
30Muse Spark 1.20.7107$36.53207.7M800k2h 20m5271
31Qwen3.8-Max0.7023$29.6470.4M1.9M19h 40m3301
32Qwen3.8-Max0.6658$53.16162.1M1.8M18h 29m4952
33Grok 4.60.6586$55.7047.5M700k4h 6m19410
34Grok 4.60.6473$57.8851.1M725k4h 13m2105
35GPT-5.60.6303$111.93116.6M261k11h 2m4253
36Kimi K30.6169$112.73307.9M980k13h 16m70831
37GPT-5.60.6093$23.5332.7M162k1h 28m2073
38DeepSeek V4 Flash Vision Exp0.4680$3.00159.1M520k6h 30m4425
39GPT-5.60.4658$119.15122.5M308k3h 17m4636
40Gemini 3.7 Flash0.3912$15.89144.8M798k2h 17m4795
41GPT-5.60.3383$180.46175.3M266k4h 7m47410
42Gemini 3.7 Flash0.3149$10.7497.9M514k1h 12m2782
43Grok 4.60.2777$27.6423.0M551k3h 18m1272
44Gemini 3.7 Flash0.2425$12.51121.4M559k2h 10m3262
45Gemini 3.7 Flash0.2101$6.7049.8M379k47m2123
46Inkling0.0000$7.507.7M51k1h 46m1351
47Muse Spark 1.20.0000$19.4092.0M1.0M4h 3m3073
48Muse Spark 1.20.0000$9.9353.1M354k2h 11m2586
49Muse Spark 1.20.0000$8.0539.0M223k1h 17m2014
50Inkling0.0000$4.664.8M18k1h 13m912

Higgs Uncertainty Inference

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.9930$40.5965.5M322k8h 40m2419
2Claude Fable 5.10.9895$33.5652.1M309k8h 2m2218
3Qwen3.8-Max0.9778$23.7479.2M556k12h 13m42814
4DeepSeek V4 Flash Vision Exp0.9597$1.5170.8M347k3h 23m2834
5Kimi K30.9593$29.5562.1M341k12h 49m28515
6Grok 4.60.9143$37.2334.8M407k3h 22m15810
7Muse Spark 1.20.9041$8.8738.4M436k2h 57m1501
8Grok 4.60.8991$42.7739.2M529k3h 38m1839
9Grok 4.60.8827$39.9237.7M462k3h 38m1818
10Inkling0.8760$6.246.3M65k37m1401
11Grok 4.60.8188$274.70264.2M1.9M19h 38m107867
12GPT-5.60.7863$99.9097.4M228k4h 26m32720
13Qwen3.8-Max0.7685$16.6651.1M571k13h 28m3425
14DeepSeek V4 Flash Vision Exp0.7308$2.17110.7M423k3h 32m3749
15Muse Spark 1.20.7187$4.4015.5M216k1h 29m1102
16Muse Spark 1.20.7132$3.9618.1M225k49m1264
17Qwen3.8-Max0.4289$16.2149.4M555k8h 58m2858
18Kimi K30.4200$32.1858.0M281k15h 50m2808
19Qwen3.8-Max0.3449$15.6147.3M543k7h 37m2788
20Claude Fable 5.10.3345$30.4545.8M311k4h 14m17415
21Qwen3.8-Max0.3078$18.5657.9M578k8h 28m3226
22Grok 4.60.2104$116.88112.5M824k9h 43m47052
23GPT-5.60.2084$52.2656.6M152k2h 42m2445
24Claude Fable 5.10.2017$30.5147.2M284k6h 48m1949
25Gemini 3.7 Flash0.1681$1.729.3M135k1h 3m1021
26GLM-5.30.1632$24.1184.1M350k8h 48m3108
27Muse Spark 1.20.1494$13.1065.0M597k2h 52m2671
28Gemini 3.7 Flash0.1295$1.9810.2M141k50m1053
29Gemini 3.7 Flash0.1075$1.629.2M146k1h 7m1021
30GLM-5.30.1005$24.4671.1M367k17h 31m2837
31Kimi K30.0960$32.3577.4M380k13h 56m31115
32Gemini 3.7 Flash0.0798$2.5310.3M161k1h 26m1101
33Claude Fable 5.10.0778$29.8844.4M316k2h 48m18313
34GPT-5.60.0777$44.3350.5M161k1h 56m2296
35DeepSeek V4 Flash Vision Exp0.0718$2.02102.8M397k15h 29m3464
36DeepSeek V4 Flash Vision Exp0.0700$1.5070.3M352k6h 13m2906
37Gemini 3.7 Flash0.0647$3.7920.7M194k1h 41m1392
38Kimi K30.0626$34.0868.6M372k14h 20m27512
39DeepSeek V4 Flash Vision Exp0.0559$1.3262.9M305k6h 32m2784
40GLM-5.30.0517$9.9130.6M263k2h 10m1675
41Muse Spark 1.20.0498$2.478.0M164k1h 0m832
42GLM-5.30.0375$14.9449.5M308k3h 28m2232
43Kimi K30.0289$90.08230.0M659k20h 0m5551
44GLM-5.30.0251$17.6860.0M356k3h 29m2463
45GPT-5.60.0080$56.4460.4M164k6h 2m25912
46Inkling0.0078$3.623.6M51k2h 38m1201
47GPT-5.60.0049$70.6365.2M124k2h 29m1946
48Inkling0.0004$3.373.3M54k1h 37m1041
49Inkling0.0004$4.084.1M54k1h 12m1122
50Inkling0.0000$2.472.5M36k1h 12m1061

SGLang Inference System Optimization

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1GPT-5.60.1092$291.23284.1M251k5h 12m61612
2GPT-5.60.0675$443.00437.3M281k6h 23m87714
3GLM-5.30.0534$68.77241.2M448k15h 36m6508
4Muse Spark 1.20.0532$132.47498.4M1.4M17h 13m14443
5Grok 4.60.0322$488.14454.2M2.3M19h 33m170383
6Qwen3.8-Max0.0214$45.80153.5M1.1M15h 59m7157
7GLM-5.30.0211$67.37244.9M383k14h 17m69820
8Gemini 3.7 Flash0.0169$19.15195.9M289k3h 33m5054
9GPT-5.60.0168$283.36279.0M247k6h 47m65521
10Kimi K30.0103$110.14311.1M507k20h 1m74011
11Muse Spark 1.20.0099$100.58403.5M1.7M11h 46m11938
12Muse Spark 1.20.0081$15.0390.2M152k1h 36m2117
13Qwen3.8-Max0.0081$33.4299.7M1.3M15h 1m5748
14Muse Spark 1.20.0047$67.55167.8M657k12h 12m5416
15Grok 4.60.0025$162.38158.0M937k10h 27m64914
16Grok 4.60.0012$313.02296.5M1.9M18h 41m112019
17Grok 4.60.0007$96.0091.3M481k3h 53m3293
18Qwen3.8-Max0.0004$47.32149.7M1.5M17h 59m6513
19GLM-5.30.0004$46.58170.2M369k10h 37m5917
20GLM-5.30.0000$50.83180.8M386k12h 52m5764
21Claude Fable 5.10.0000$219.62392.8M963k7h 58m8449
22Muse Spark 1.20.0000$32.8165.0M127k4h 50m2763
23Claude Fable 5.10.0000$193.70342.2M936k9h 59m82011
24DeepSeek V4 Flash Vision Exp0.0000$2.44138.4M299k9h 52m5314
25DeepSeek V4 Flash Vision Exp0.0000$5.46332.4M492k10h 53m78215
26DeepSeek V4 Flash Vision Exp0.0000$3.59206.9M443k9h 31m59316
27Kimi K30.0000$185.83517.0M1.5M20h 0m118813
28GPT-5.60.0000$109.76114.5M139k5h 57m4017
29Inkling0.0000$4.875.1M12k1h 40m923
30Claude Fable 5.10.0000$177.26316.2M794k11h 11m8339
31Kimi K30.0000$246.45710.4M1.8M20h 0m173614
32GLM-5.30.0000$68.85206.5M390k14h 23m60717
33Inkling0.0000$7.608.0M10k1h 46m1202
34Gemini 3.7 Flash0.0000$9.47100.2M149k1h 43m3164
35GPT-5.60.0000$182.18175.0M189k6h 4m4858
36Inkling0.0000$5.185.4M12k1h 28m912
37DeepSeek V4 Flash Vision Exp0.0000$3.41198.5M360k8h 1m5624
38Grok 4.60.0000$419.76405.9M2.2M19h 4m158660
39Claude Fable 5.10.0000$182.53329.2M746k9h 31m84121
40Inkling0.0000$6.426.7M16k1h 12m1001
41DeepSeek V4 Flash Vision Exp0.0000$3.54209.2M364k8h 32m6329
42Qwen3.8-Max0.0000$42.20143.0M952k17h 51m6041
43Kimi K30.0000$49.64141.5M328k6h 19m5297
44Inkling0.0000$14.7615.4M36k34m1942
45Gemini 3.7 Flash0.0000$9.0285.7M143k1h 52m3061
46Qwen3.8-Max0.0000$47.25152.8M1.4M17h 10m6423
47Gemini 3.7 Flash0.0000$12.5890.8M182k4h 36m3171
48Kimi K30.0000$124.72355.5M911k15h 41m94824
49Claude Fable 5.10.0000$226.14406.3M953k11h 14m90215
50Gemini 3.7 Flash0.0000$6.5851.9M91k1h 41m2964

Lean 4 Kernel Type Checker in Pascal

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.11.0000$34.1749.9M378k2h 37m1646
2Claude Fable 5.11.0000$45.3070.2M418k4h 0m2009
3Claude Fable 5.11.0000$34.2250.9M360k1h 59m1598
4Claude Fable 5.11.0000$43.0867.7M379k4h 3m20611
5Claude Fable 5.11.0000$50.9183.1M385k3h 9m2328
6GPT-5.60.9807$221.16222.7M429k6h 13m69119
7GPT-5.60.9791$78.0186.8M272k4h 32m3952
8GPT-5.60.9757$207.31199.1M381k7h 23m62010
9Gemini 3.7 Flash0.9448$95.08678.0M1.4M19h 49m16601
10Gemini 3.7 Flash0.9361$54.83453.2M1.3M12h 43m12241
11GPT-5.60.9339$340.15337.0M382k8h 35m78911
12Kimi K30.9198$192.68537.9M1.4M20h 0m12181
13GLM-5.30.8650$139.71481.8M637k18h 8m8596
14GLM-5.30.8372$111.12401.3M867k19h 42m90410
15Kimi K30.8313$155.97405.9M1.5M19h 47m9353
16GPT-5.60.7874$397.16404.0M396k8h 54m11347
17Kimi K30.5265$147.65387.7M1.4M20h 2m8941
18Kimi K30.4229$171.33477.1M1.3M19h 55m113012
19GLM-5.30.4065$133.87385.6M660k19h 35m6773
20Grok 4.60.3824$105.13105.0M1.0M6h 34m4902
21DeepSeek V4 Flash Vision Exp0.3599$12.71725.7M1.6M10h 38m16144
22DeepSeek V4 Flash Vision Exp0.3164$9.48551.4M1.2M7h 58m12288
23Kimi K30.2644$151.43423.8M1.1M20h 1m9281
24DeepSeek V4 Flash Vision Exp0.2513$13.44768.2M1.8M13h 43m17607
25Gemini 3.7 Flash0.1248$74.81557.5M1.1M14h 22m12171
26GLM-5.30.1056$134.85474.9M964k19h 7m11053
27DeepSeek V4 Flash Vision Exp0.0143$12.56761.2M1.2M12h 33m162735
28Gemini 3.7 Flash0.0087$48.94384.1M939k13h 20m11181
29Muse Spark 1.20.0000$19.14108.5M529k2h 29m3481
30Qwen3.8-Max0.0000$14.1428.2M1.1M19h 59m1691
31Inkling0.0000$5.565.7M37k1h 16m1152
32Inkling0.0000$5.465.6M34k55m1072
33Muse Spark 1.20.0000$21.0953.6M699k3h 48m2052
34Muse Spark 1.20.0000$29.46162.8M830k6h 1m4043
35Qwen3.8-Max0.0000$8.3110.6M914k19h 59m1092
36Inkling0.0000$5.705.9M22k14m802
37Grok 4.60.0000$114.14110.5M1.1M7h 47m4924
38Muse Spark 1.20.0000$27.3088.9M859k4h 46m2781
39Grok 4.60.0000$407.54406.7M2.0M19h 47m1527137
40Grok 4.60.0000$250.10240.1M1.8M13h 13m97043
41Inkling0.0000$13.8114.4M43k1h 26m1751
42DeepSeek V4 Flash Vision Exp0.0000$11.64694.6M1.3M11h 22m141333
43Grok 4.60.0000$233.84217.8M2.5M19h 53m92930
44Muse Spark 1.20.0000$9.5148.0M382k2h 35m2132
45Gemini 3.7 Flash0.0000$101.50861.0M1.8M19h 58m20921
46Qwen3.8-Max0.0000$13.9225.5M1.2M19h 56m1672
47Qwen3.8-Max0.0000$19.6845.8M1.3M20h 0m2111
48GLM-5.30.0000$113.59400.7M1.0M18h 31m9625
49Qwen3.8-Max0.0000$10.3513.6M1.1M19h 57m1151
50Inkling0.0000$6.566.8M25k12m1096

libexpat Optimization

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.5674$350.25628.4M1.5M16h 3m145682
2Claude Fable 5.10.5395$313.80554.2M1.5M11h 26m128934
3Claude Fable 5.10.4885$375.68676.2M1.5M17h 3m150550
4Claude Fable 5.10.4469$289.07503.7M1.5M13h 41m115650
5GPT-5.60.3847$146.13140.6M311k2h 25m4397
6GPT-5.60.3225$67.0675.3M223k1h 44m3269
7GPT-5.60.3025$18.3526.1M127k53m1444
8Claude Fable 5.10.2880$179.04308.9M1.0M10h 29m75512
9GPT-5.60.2868$49.1153.9M193k1h 32m2385
10GPT-5.60.2751$110.05104.7M243k2h 24m3507
11Gemini 3.7 Flash0.2399$41.35394.7M960k4h 30m10096
12GLM-5.30.2372$181.79673.8M1.1M8h 14m15013
13GLM-5.30.2303$183.58670.1M1.1M9h 50m146018
14GLM-5.30.1823$327.581209.3M1.8M18h 23m263546
15Kimi K30.1778$238.01679.3M1.8M19h 52m158883
16GLM-5.30.1583$236.28865.5M1.6M18h 53m199134
17Gemini 3.7 Flash0.1572$41.30358.7M785k6h 1m8995
18Gemini 3.7 Flash0.1555$24.69216.5M877k2h 56m6242
19Kimi K30.1513$169.56484.2M1.2M19h 34m115516
20Grok 4.60.1479$150.76148.7M804k4h 11m58026
21Grok 4.60.1393$108.10108.1M736k4h 5m43511
22Kimi K30.1288$173.14509.0M1.1M13h 23m114928
23Grok 4.60.1022$79.4275.8M516k2h 56m2737
24DeepSeek V4 Flash Vision Exp0.0891$17.041034.3M1.6M6h 42m22736
25DeepSeek V4 Flash Vision Exp0.0545$16.46992.0M1.7M6h 43m222211
26Muse Spark 1.20.0000$40.00150.5M651k3h 23m3883
27Muse Spark 1.20.0000$41.66132.7M956k3h 49m4733
28Inkling0.0000$23.4724.5M46k39m3592
29Muse Spark 1.20.0000$16.5487.9M378k1h 23m3093
30Qwen3.8-Max0.0000$21.1536.8M1.9M20h 0m260
31Inkling0.0000$3.703.7M39k18m852
32Qwen3.8-Max0.0000$16.2822.3M1.7M19h 46m202
33Gemini 3.7 Flash0.0000$33.92328.7M960k2h 58m81020
34DeepSeek V4 Flash Vision Exp0.0000$13.51827.8M1.3M5h 43m172013
35DeepSeek V4 Flash Vision Exp0.0000$21.931323.6M2.2M8h 50m30957
36Gemini 3.7 Flash0.0000$29.56325.1M599k2h 10m8302
37DeepSeek V4 Flash Vision Exp0.0000$33.361965.0M3.8M19h 31m410248
38Kimi K30.0000$214.44584.8M1.8M20h 0m14591
39Inkling0.0000$18.3619.0M75k47m2514
40Qwen3.8-Max0.0000$19.2333.9M1.7M19h 53m2331
41Grok 4.60.0000$93.7799.5M677k3h 26m43510
42Muse Spark 1.20.0000$39.83172.4M582k3h 18m5596
43Inkling0.0000$3.583.6M31k16m742
44Qwen3.8-Max0.0000$25.1246.8M2.2M19h 55m301
45Muse Spark 1.20.0000$32.92157.9M806k3h 52m3443
46Grok 4.60.0000$71.7870.3M399k1h 58m2775
47Qwen3.8-Max0.0000$13.0614.1M1.6M19h 35m156
48Inkling0.0000$3.113.2M24k13m823
49Kimi K30.0000$256.75739.1M1.4M20h 0m15761
50GLM-5.30.0000$216.67774.3M1.6M16h 7m175419

Lua Native Compiler

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.11.0000$44.1969.8M381k3h 24m2158
2Claude Fable 5.11.0000$51.0382.5M402k3h 56m24512
3Claude Fable 5.11.0000$55.8488.8M470k3h 14m2356
4GLM-5.30.9976$201.34735.2M1.4M11h 37m16603
5GLM-5.30.9953$142.13511.6M1.2M9h 22m11782
6DeepSeek V4 Flash Vision Exp0.9788$23.061383.4M2.4M13h 19m291121
7GLM-5.30.9599$274.761000.1M1.8M19h 32m22735
8Kimi K30.9575$183.51524.3M1.3M19h 59m120320
9GLM-5.30.9527$236.30883.7M1.1M17h 20m18357
10Kimi K30.8750$185.22530.7M1.3M20h 0m12071
11Claude Fable 5.10.8463$43.7869.0M381k3h 39m2311
12Gemini 3.7 Flash0.6998$21.69183.4M580k2h 58m4001
13Claude Fable 5.10.6667$57.3892.1M466k4h 40m2487
14Gemini 3.7 Flash0.6541$62.39519.3M1.2M7h 8m12601
15Gemini 3.7 Flash0.6517$26.08255.3M824k1h 58m6701
16Gemini 3.7 Flash0.6392$29.13278.2M822k3h 5m7511
17Grok 4.60.6384$176.45167.1M1.5M7h 26m6538
18Kimi K30.5063$220.93645.6M1.2M20h 2m14611
19Gemini 3.7 Flash0.3270$79.34743.6M1.7M8h 42m18151
20Kimi K30.2862$159.88431.8M1.4M19h 11m10091
21DeepSeek V4 Flash Vision Exp0.2005$21.271250.2M2.4M15h 16m267617
22Inkling0.0000$27.7428.9M72k42m3001
23Inkling0.0000$16.9217.5M67k44m2211
24Qwen3.8-Max0.0000$64.51206.8M2.0M19h 8m6392
25DeepSeek V4 Flash Vision Exp0.0000$11.39676.4M1.3M7h 45m150715
26Grok 4.60.0000$110.36107.0M986k5h 11m4478
27Inkling0.0000$48.6250.6M132k1h 1m3672
28GPT-5.60.0000$35.0744.0M172k2h 41m21110
29Muse Spark 1.20.0000$44.95118.1M616k6h 24m3542
30Kimi K30.0000$228.55670.7M1.4M19h 8m14083
31Inkling0.0000$14.8115.3M75k53m2231
32GPT-5.60.0000$7.458.9M65k25m653
33GLM-5.30.0000$251.10909.6M1.3M19h 55m19186
34Qwen3.8-Max0.0000$121.10426.7M2.2M19h 59m9841
35Grok 4.60.0000$271.55264.4M2.0M9h 49m106131
36Muse Spark 1.20.0000$10.7739.0M388k1h 41m1644
37DeepSeek V4 Flash Vision Exp0.0000$22.431385.4M2.0M11h 31m28602
38Qwen3.8-Max0.0000$135.25480.8M2.3M19h 7m10812
39Qwen3.8-Max0.0000$60.45189.1M2.1M19h 48m5993
40Qwen3.8-Max0.0000$31.0087.6M1.4M13h 21m3856
41Inkling0.0000$7.918.2M23k26m1652
42Grok 4.60.0000$676.54663.0M4.1M19h 59m2429277
43GPT-5.60.0000$127.72124.4M217k2h 5m38224
44Muse Spark 1.20.0000$4.4622.6M149k52m1452
45DeepSeek V4 Flash Vision Exp0.0000$18.341059.5M2.2M17h 1m23839
46GPT-5.60.0000$24.8437.1M163k1h 9m21913
47Grok 4.60.0000$250.76240.6M2.1M11h 13m93016
48Muse Spark 1.20.0000$22.61124.4M368k1h 49m3232
49Muse Spark 1.20.0000$12.0053.2M328k1h 52m1892
50GPT-5.60.0000$20.5039.9M168k1h 9m2317

† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.

Machine-Learned Interatomic Potential

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1GPT-5.60.9925$108.64109.2M151k11h 16m3488
2GPT-5.60.9918$109.29114.5M152k15h 47m40416
3GLM-5.30.9901$21.3550.8M160k19h 5m3531
4Qwen3.8-Max0.9731$4.1011.4M180k17h 54m1853
5GLM-5.30.9615$15.6237.8M156k19h 29m2942
6Claude Fable 5.10.9586$14.2221.7M140k16h 34m17911
7GPT-5.60.9437$26.2041.3M108k14h 19m2373
8Claude Fable 5.10.9369$12.1118.4M119k17h 11m17515
9Kimi K30.9282$14.8121.6M194k19h 20m1753
10Qwen3.8-Max0.9220$5.1315.3M180k19h 14m2092
11GLM-5.30.9184$31.3162.0M243k18h 46m3072
12Kimi K30.9177$11.7220.5M161k18h 47m1653
13DeepSeek V4 Flash Vision Exp0.9173$3.12179.4M413k18h 25m5586
14Claude Fable 5.10.9113$11.2616.5M124k17h 50m1505
15Grok 4.60.8964$27.9418.3M258k19h 42m1472
16Kimi K30.8916$6.549.0M95k18h 28m1262
17Muse Spark 1.20.8888$4.0820.1M191k2h 11m16512
18Claude Fable 5.10.8820$17.0526.5M157k16h 32m1797
19Grok 4.60.8799$71.4258.5M227k19h 56m3079
20Muse Spark 1.20.8642$31.30153.8M371k2h 47m46412
21Gemini 3.7 Flash0.8422$11.4530.2M159k8h 5m2083
22Grok 4.60.8342$80.8855.9M251k19h 37m26510
23Claude Fable 5.10.8229$11.9918.2M119k18h 9m17113
24Muse Spark 1.20.8223$3.4515.9M178k49m1273
25GLM-5.30.8163$20.3255.4M207k19h 30m3502
26Qwen3.8-Max0.8093$6.9420.8M251k18h 48m2312
27Kimi K30.8033$15.1121.0M176k19h 11m1683
28GLM-5.30.7858$25.2849.5M200k18h 31m2852
29Muse Spark 1.20.7533$3.6517.8M164k1h 53m1263
30Kimi K30.7378$9.8813.2M119k19h 28m1432
31Gemini 3.7 Flash0.7285$10.9529.5M154k8h 16m2137
32Grok 4.60.7267$28.2320.6M259k19h 44m1593
33Muse Spark 1.20.7264$3.5517.0M170k2h 17m1443
34DeepSeek V4 Flash Vision Exp0.6906$1.8093.9M332k18h 0m3786
35Grok 4.60.5686$61.5946.3M291k18h 29m2504
36Gemini 3.7 Flash0.5113$8.2127.8M178k5h 29m1842
37DeepSeek V4 Flash Vision Exp0.4712$1.0554.4M192k19h 29m3422
38Gemini 3.7 Flash0.4203$5.2521.8M116k4h 45m1781
39DeepSeek V4 Flash Vision Exp0.2438$1.5984.5M269k19h 1m3895
40Qwen3.8-Max0.2071$6.2316.9M295k18h 26m2182
41Inkling0.0572$3.463.5M36k2h 37m1361
42DeepSeek V4 Flash Vision Exp0.0562$2.23120.5M374k18h 41m4363
43Inkling0.0431$4.945.0M59k45m1201
44Inkling0.0393$6.847.0M43k1h 26m1811
45Inkling0.0179$5.355.4M56k1h 32m1262
46Inkling0.0126$4.985.1M42k1h 33m1482
47GPT-5.60.0076$48.9956.5M122k14h 29m2935
48GPT-5.60.0010$36.5549.8M109k7h 20m2723
49Qwen3.8-Max0.0010$4.4112.0M206k19h 18m1791
50Gemini 3.7 Flash0.0010$6.0120.4M174k4h 35m1542

Medium-Range Weather Forecast

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.6214$15.1124.2M125k16h 58m20411
2Claude Fable 5.10.6178$12.6920.4M104k18h 1m17512
3Claude Fable 5.10.6011$16.8127.4M129k16h 5m20816
4Claude Fable 5.10.5960$16.0625.6M135k16h 36m19312
5GPT-5.60.5936$28.8247.8M113k15h 13m2944
6GLM-5.30.5873$22.6171.9M228k18h 20m3142
7Grok 4.60.5804$120.40104.0M444k19h 13m46437
8Gemini 3.7 Flash0.5747$22.4049.9M145k13h 34m2567
9Gemini 3.7 Flash0.5566$16.5654.7M160k9h 58m2852
10Grok 4.60.5522$208.02184.0M911k19h 55m82653
11Grok 4.60.5495$238.57230.6M626k20h 0m86162
12Gemini 3.7 Flash0.5365$13.1137.7M147k6h 21m2413
13Gemini 3.7 Flash0.5356$19.3648.1M172k8h 5m2524
14Gemini 3.7 Flash0.5332$4.9921.4M136k4h 20m1711
15GLM-5.30.4818$19.7547.5M201k18h 1m2735
16GLM-5.30.4745$28.0396.3M327k19h 20m3637
17GLM-5.30.4723$21.0347.8M206k18h 36m2932
18GLM-5.30.4679$26.3077.9M252k19h 29m3642
19DeepSeek V4 Flash Vision Exp0.4639$1.5583.0M253k19h 5m4035
20Grok 4.60.4530$43.0735.9M245k19h 35m2251
21Grok 4.60.4154$92.4866.7M263k19h 27m3036
22Qwen3.8-Max0.4005$8.9429.1M230k19h 22m2811
23Qwen3.8-Max0.3989$14.9148.7M387k19h 25m3571
24Qwen3.8-Max0.3606$13.1244.7M268k19h 18m3853
25Muse Spark 1.20.3456$4.3818.4M145k37m1042
26Qwen3.8-Max0.3410$4.9114.6M172k18h 25m1894
27Qwen3.8-Max0.3360$8.0125.7M213k18h 56m2398
28Muse Spark 1.20.3227$38.37235.4M385k2h 26m5925
29Muse Spark 1.20.3018$11.6245.2M202k1h 9m1817
30Muse Spark 1.20.2944$2.128.7M99k50m861
31Muse Spark 1.20.2758$11.1751.5M378k2h 40m2116
32Inkling0.2568$3.543.6M37k1h 43m971
33Inkling0.2237$3.913.9M47k1h 43m1201
34Inkling0.1542$5.245.4M39k1h 50m1322
35Inkling0.1482$3.493.6M19k1h 32m1021
36GPT-5.60.0000$122.35137.1M209k17h 41m5767
37Kimi K30.0000$20.2549.9M257k19h 31m2968
38Kimi K30.0000$16.0342.3M172k19h 35m2821
39Kimi K30.0000$13.2932.3M193k19h 22m2452
40Inkling0.0000$6.206.2M72k38m1321
41DeepSeek V4 Flash Vision Exp0.0000$1.6690.1M258k19h 29m3923
42DeepSeek V4 Flash Vision Exp0.0000$1.99110.4M289k18h 13m4874
43DeepSeek V4 Flash Vision Exp0.0000$1.5584.2M243k18h 42m3985
44GPT-5.60.0000$63.6270.4M130k17h 56m2903
45Kimi K30.0000$11.0428.1M135k19h 31m2616
46GPT-5.60.0000$72.5794.1M189k15h 53m47814
47DeepSeek V4 Flash Vision Exp0.0000$1.6893.8M242k19h 27m4522
48Claude Fable 5.10.0000$14.6023.4M121k14h 28m1868
49GPT-5.60.0000$88.96100.2M138k14h 52m4168
50Kimi K30.0000$19.0948.5M196k18h 52m3266

MEG Speech Decoding

task page →43 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.3166$19.4430.2M179k16h 28m2148
2Claude Fable 5.10.3123$26.3542.8M204k16h 51m24012
3GPT-5.60.3119$41.2143.5M133k5h 23m2042
4Claude Fable 5.10.3116$13.9921.5M133k17h 11m19211
5Claude Fable 5.10.3057$16.5726.2M144k13h 23m20211
6Grok 4.60.3041$128.46122.4M984k15h 5m54523
7GPT-5.60.3039$14.1019.5M85k2h 25m1372
8GPT-5.60.2311$153.75128.1M180k10h 45m29817
9Grok 4.60.2265$217.69204.9M1.2M19h 25m87823
10GLM-5.30.2244$22.6851.0M188k18h 6m3152
11Claude Fable 5.10.2155$25.9841.7M212k14h 56m23710
12Grok 4.60.2107$132.86129.3M708k19h 25m57133
13GLM-5.30.2036$34.9674.2M244k14h 55m3722
14GPT-5.60.1902$64.3567.7M163k7h 2m2794
15GPT-5.60.1862$104.8098.3M165k7h 53m3188
16Grok 4.60.1844$95.1683.3M284k18h 58m36211
17DeepSeek V4 Flash Vision Exp0.1749$1.3268.7M240k18h 36m3596
18Muse Spark 1.20.1659$17.1036.3M144k3h 7m1306
19GLM-5.30.1492$27.3490.5M318k15h 42m3713
20Qwen3.8-Max0.1460$13.1843.8M302k19h 37m3337
21Grok 4.60.1245$71.3867.2M365k9h 53m29310
22Muse Spark 1.20.1175$9.5247.7M272k3h 44m1838
23Gemini 3.7 Flash0.1130$4.8733.1M132k3h 51m2261
24Gemini 3.7 Flash0.1121$12.1251.3M122k7h 24m3124
25Kimi K30.1102$14.9032.2M173k18h 22m2543
26GLM-5.30.1032$33.8077.7M262k19h 7m3532
27Muse Spark 1.20.1011$11.9661.9M415k4h 37m2413
28Qwen3.8-Max0.0985$20.4770.2M412k18h 53m4782
29Gemini 3.7 Flash0.0948$9.9338.7M111k7h 46m2665
30Gemini 3.7 Flash0.0854$10.1135.8M110k7h 53m3031
31DeepSeek V4 Flash Vision Exp0.0826$1.7393.5M273k12h 25m4572
32GLM-5.30.0000$49.90134.3M340k19h 4m46921
33Qwen3.8-Max0.0000$13.8445.7M331k19h 21m3362
34Inkling0.0000$5.295.5M26k1h 28m1511
35Qwen3.8-Max0.0000$27.9699.7M412k19h 32m5274
36Muse Spark 1.20.0000$13.1177.0M195k5h 0m2023
37Inkling0.0000$5.015.2M28k1h 29m1281
38Gemini 3.7 Flash0.0000$12.2048.7M173k7h 50m2852
39Muse Spark 1.20.0000$16.3673.7M275k2h 42m2447
40Inkling0.0000$6.236.3M53k2h 6m1391
41Qwen3.8-Max0.0000$8.5526.2M281k19h 56m2421
42Inkling0.0000$5.315.5M23k2h 3m1342
43Inkling0.0000$5.425.6M24k1h 30m1161

Wan 2.1 on MAX/Mojo

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.9331$62.26106.8M367k8h 14m2848
2Claude Fable 5.10.9314$52.8790.3M319k5h 49m2749
3Kimi K30.9307$34.4290.2M306k9h 26m3065
4GPT-5.60.9241$103.11107.4M168k5h 28m3698
5Claude Fable 5.10.9205$68.80118.8M390k9h 16m3236
6Kimi K30.9196$29.2075.3M345k8h 0m2868
7Claude Fable 5.10.9192$42.8772.2M281k5h 13m2265
8Claude Fable 5.10.9186$55.5395.2M329k8h 32m27910
9GLM-5.30.9144$51.16175.7M465k14h 14m4612
10GLM-5.30.9094$42.10140.4M401k16h 39m4042
11Kimi K30.9090$15.0637.9M188k7h 2m2047
12GLM-5.30.9087$43.47155.3M374k16h 47m4782
13GPT-5.60.9087$34.0444.6M108k4h 5m1999
14Grok 4.60.9085$149.23138.7M1.3M12h 2m5862
15GPT-5.60.9062$69.5278.1M124k3h 50m2953
16Grok 4.60.9047$29.2330.9M270k3h 12m1581
17GPT-5.60.9014$57.1166.4M129k4h 8m2675
18Grok 4.60.9010$76.5166.4M639k9h 42m2732
19Kimi K30.8899$31.5184.6M291k12h 22m3483
20GPT-5.60.8681$66.0672.8M105k4h 8m2674
21Qwen3.8-Max0.4407$34.5198.7M1.3M19h 41m4232
22Qwen3.8-Max0.4404$41.1995.2M1.6M20h 1m4271
23DeepSeek V4 Flash Vision Exp0.4396$4.92297.3M464k19h 11m68111
24DeepSeek V4 Flash Vision Exp0.4384$3.53204.1M419k17h 55m5623
25DeepSeek V4 Flash Vision Exp0.3550$2.99168.7M386k19h 32m4533
26Qwen3.8-Max0.2981$45.25145.8M997k19h 53m5261
27Kimi K30.2838$85.29208.6M954k19h 56m5583
28DeepSeek V4 Flash Vision Exp0.2720$3.61208.7M436k14h 28m56211
29Grok 4.60.2716$127.41108.5M1.7M19h 30m4644
30DeepSeek V4 Flash Vision Exp0.2506$3.27186.8M413k17h 5m4935
31Grok 4.60.2296$214.90198.0M1.7M19h 41m75913
32GLM-5.30.0945$60.08134.7M351k20h 1m4321
33Qwen3.8-Max0.0664$30.3485.4M1.2M19h 11m3602
34Inkling0.0429$17.1217.9M30k2h 11m1921
35Inkling0.0404$13.8514.5M23k1h 11m1421
36Gemini 3.7 Flash0.0349$23.63178.6M776k6h 26m5691
37Gemini 3.7 Flash0.0347$27.48166.6M453k6h 47m4381
38Gemini 3.7 Flash0.0331$42.63248.9M797k11h 49m7161
39Muse Spark 1.20.0325$10.9349.7M405k2h 44m1823
40Gemini 3.7 Flash0.0317$28.92149.4M462k8h 34m3801
41Gemini 3.7 Flash0.0310$36.97183.1M560k8h 49m4831
42Inkling0.0305$12.9313.5M35k14m1411
43Muse Spark 1.20.0252$38.37181.5M771k3h 21m48113
44GLM-5.30.0239$77.39173.5M511k19h 46m46411
45Qwen3.8-Max0.0000$50.91123.3M1.9M19h 56m4671
46Muse Spark 1.20.0000$29.55111.8M325k3h 53m3312
47Muse Spark 1.20.0000$16.0877.6M385k3h 9m2542
48Inkling0.0000$16.0516.8M29k1h 23m1762
49Muse Spark 1.20.0000$10.8049.7M240k3h 10m2263
50Inkling0.0000$11.4711.9M30k1h 18m1452

† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.

MS/MS De Novo Generation

task page →45 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1GPT-5.60.1316$55.6569.9M170k12h 29m3322
2GPT-5.60.1312$87.71101.7M199k15h 19m4536
3Kimi K30.1101$21.3546.7M229k17h 35m2826
4Kimi K30.1097$26.9062.8M258k19h 1m3257
5Kimi K30.1091$18.2139.5M209k19h 3m2395
6Gemini 3.7 Flash0.1064$8.8840.2M172k5h 7m2652
7GLM-5.30.1050$34.1375.8M305k18h 42m3362
8GLM-5.30.1036$36.9494.3M365k18h 38m3593
9Grok 4.60.1023$289.71280.9M1.3M20h 0m110991
10GLM-5.30.0941$44.44143.9M417k17h 20m4508
11Qwen3.8-Max0.0879$12.0738.5M344k19h 44m2903
12Gemini 3.7 Flash0.0865$5.2520.8M119k3h 41m1721
13Gemini 3.7 Flash0.0846$10.2648.6M211k7h 0m2643
14Grok 4.60.0709$86.3377.3M437k10h 44m3176
15Qwen3.8-Max0.0652$23.1077.1M550k18h 46m4362
16Muse Spark 1.20.0557$16.8190.3M275k5h 28m3424
17Muse Spark 1.20.0507$33.64201.0M350k2h 20m4883
18Inkling0.0000$4.784.9M42k32m1361
19GLM-5.30.0000$28.5571.5M295k19h 12m3322
20GLM-5.30.0000$43.34125.4M382k18h 56m4632
21GPT-5.60.0000$29.4443.7M172k11h 11m2914
22Inkling0.0000$9.609.7M91k1h 37m1871
23Qwen3.8-Max0.0000$18.8059.9M558k19h 4m3661
24DeepSeek V4 Flash Vision Exp0.0000$1.89103.3M298k19h 46m4353
25DeepSeek V4 Flash Vision Exp0.0000$2.64142.6M438k12h 9m4745
26Inkling0.0000$4.224.3M40k1h 20m1252
27Muse Spark 1.20.0000$8.1142.2M266k2h 10m1692
28Qwen3.8-Max0.0000$22.7775.2M580k19h 13m4923
29Grok 4.60.0000$51.2439.8M426k15h 47m2435
30Kimi K30.0000$18.6638.3M209k20h 0m2544
31DeepSeek V4 Flash Vision Exp0.0000$2.54140.9M373k17h 33m4735
32Qwen3.8-Max0.0000$31.11106.6M635k17h 29m5003
33Inkling0.0000$9.219.4M67k2h 19m2351
34GPT-5.60.0000$115.28137.3M233k11h 7m60310
35Kimi K30.0000$31.9877.8M334k19h 25m3394
36DeepSeek V4 Flash Vision Exp0.0000$2.76151.1M447k15h 16m5036
37GPT-5.60.0000$33.4341.9M177k7h 37m2574
38Gemini 3.7 Flash0.0000$6.9328.3M150k3h 32m2034
39Grok 4.60.0000$192.75185.2M1.2M14h 10m76664
40DeepSeek V4 Flash Vision Exp0.0000$2.75153.9M409k18h 4m5102
41Grok 4.60.0000$86.8176.6M468k16h 53m3369
42Gemini 3.7 Flash0.0000$15.5862.6M304k13h 22m2852
43Muse Spark 1.20.0000$7.4939.5M207k4h 45m2182
44Muse Spark 1.20.0000$9.6433.9M298k3h 46m1945
45Inkling0.0000$6.196.3M42k29m1201

Multi-GPU Efficient Finetuning

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Inkling0.5238$11.7812.1M59k1h 35m1661
2Muse Spark 1.20.3810$64.94126.7M662k16h 40m38627
3DeepSeek V4 Flash Vision Exp0.3333$0.8236.7M207k19h 58m2381
4Kimi K30.3333$19.1518.0M131k19h 9m1676
5Muse Spark 1.20.3095$11.9541.8M358k12h 44m19924
6Kimi K30.3095$16.5615.3M148k19h 10m1343
7GLM-5.30.2857$12.8826.8M144k19h 0m2342
8Claude Fable 5.10.2857$5.657.6M75k17h 4m982
9Qwen3.8-Max0.2857$6.4217.1M316k18h 44m2135
10Claude Fable 5.10.2857$5.387.9M60k16h 16m1183
11GLM-5.30.2857$10.4218.4M127k19h 10m1756
12Muse Spark 1.20.2619$355.312070.9M1.9M15h 46m592022
13Claude Fable 5.10.2619$6.128.6M75k18h 44m1124
14Qwen3.8-Max0.2619$6.1516.6M296k17h 38m22010
15GPT-5.60.2381$19.3323.5M110k15h 47m1673
16Claude Fable 5.10.2143$5.267.1M71k18h 19m923
17Gemini 3.7 Flash0.2143$10.2429.6M72k9h 49m18131
18Muse Spark 1.20.1905$5.5220.2M214k2h 15m1353
19Kimi K30.1905$16.1311.8M128k18h 47m1161
20Gemini 3.7 Flash0.1667$11.8531.1M102k19h 53m20210
21Claude Fable 5.10.1667$6.929.6M88k17h 1m1014
22DeepSeek V4 Flash Vision Exp0.1429$0.9241.7M226k19h 50m2512
23GLM-5.30.0714$17.6133.2M221k18h 56m2112
24Muse Spark 1.20.0476$35.64132.6M581k10h 53m45113
25Kimi K30.0476$25.1335.1M233k19h 37m2361
26Inkling0.0476$5.355.5M24k1h 26m1322
27GPT-5.60.0238$19.6525.0M71k12h 53m1333
28Inkling0.0238$14.2214.8M45k2h 19m2191
29DeepSeek V4 Flash Vision Exp0.0238$0.6629.1M168k19h 41m2262
30Qwen3.8-Max0.0000$4.6511.0M287k19h 49m1835
31Gemini 3.7 Flash0.0000$22.1847.8M107k18h 31m2992
32GPT-5.60.0000$16.4525.9M78k16h 50m1802
33GLM-5.30.0000$20.2634.9M208k19h 31m2114
34Inkling0.0000$6.666.8M46k2h 1m1351
35GPT-5.60.0000$24.1334.7M95k16h 10m1812
36GPT-5.60.0000$30.4139.1M90k14h 8m1993
37Grok 4.60.0000$47.0350.4M373k19h 22m26113
38Qwen3.8-Max0.0000$3.679.9M171k19h 13m1533
39GLM-5.30.0000$17.6732.5M199k19h 29m2152
40DeepSeek V4 Flash Vision Exp0.0000$0.7637.0M151k19h 59m2751
41Qwen3.8-Max0.0000$7.4421.5M284k19h 29m2532
42Grok 4.60.0000$18.5324.3M327k17h 14m1947
43DeepSeek V4 Flash Vision Exp0.0000$1.2360.8M248k19h 49m3231
44Grok 4.60.0000$32.8828.8M249k18h 12m1965
45Gemini 3.7 Flash0.0000$8.3027.1M102k9h 38m18810
46Grok 4.60.0000$23.9518.0M271k19h 15m1454
47Inkling0.0000$4.164.3M15k34m1073
48Kimi K30.0000$10.948.7M101k18h 40m1211
49Gemini 3.7 Flash0.0000$30.3173.0M132k18h 49m31013
50Grok 4.60.0000$20.0718.4M217k18h 24m1568

Notebook Compression

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.6002$67.85117.2M379k14h 26m40117
2Claude Fable 5.10.5736$77.68134.5M428k13h 49m45016
3Claude Fable 5.10.5457$68.00118.2M365k13h 15m41921
4Claude Fable 5.10.5279$64.70113.1M334k17h 24m41917
5Claude Fable 5.10.4824$61.77107.1M338k15h 49m41816
6GLM-5.30.2770$25.8686.5M327k19h 49m3703
7GLM-5.30.2744$34.96116.4M391k19h 2m4352
8Kimi K30.2647$37.6696.4M363k19h 31m40917
9GPT-5.60.2623$123.00167.4M279k11h 19m6288
10GPT-5.60.2599$99.01115.0M257k15h 3m52810
11GPT-5.60.2438$58.4876.1M190k17h 55m41214
12Kimi K30.2338$43.76111.9M405k17h 35m40312
13GPT-5.60.2326$88.19110.8M211k11h 35m5186
14Grok 4.60.2286$85.3184.1M498k7h 48m33114
15Kimi K30.2242$36.2787.9M412k11h 52m38212
16Qwen3.8-Max0.2190$26.3283.3M818k19h 14m4542
17Qwen3.8-Max0.2153$10.8227.5M610k19h 5m3142
18Kimi K30.2151$24.4560.7M274k11h 45m30210
19Grok 4.60.2146$36.9834.6M412k13h 20m2205
20DeepSeek V4 Flash Vision Exp0.2139$1.97101.8M387k13h 26m4201
21Qwen3.8-Max0.2125$16.1547.4M646k17h 46m3622
22Grok 4.60.2098$96.3888.3M636k11h 39m37712
23GLM-5.30.2089$39.2889.8M362k20h 0m3691
24Grok 4.60.2060$71.5475.2M388k6h 34m38127
25GLM-5.30.2058$59.65209.0M565k15h 11m5488
26DeepSeek V4 Flash Vision Exp0.2028$2.58144.1M374k19h 46m5082
27Grok 4.60.2019$87.9283.6M470k8h 37m36110
28Qwen3.8-Max0.2005$15.4343.7M693k18h 44m4141
29Kimi K30.1961$13.5921.1M194k9h 53m1799
30DeepSeek V4 Flash Vision Exp0.1948$3.01169.4M449k11h 0m5514
31DeepSeek V4 Flash Vision Exp0.1939$2.93164.0M435k14h 31m5264
32Qwen3.8-Max0.1831$20.1858.6M851k19h 8m4432
33DeepSeek V4 Flash Vision Exp0.1818$2.40126.7M415k10h 56m4216
34Gemini 3.7 Flash0.1775$2.9513.6M108k3h 17m1292
35Gemini 3.7 Flash0.1714$6.9227.7M160k6h 28m2365
36Muse Spark 1.20.1646$89.68227.7M1.2M20h 2m7287
37GPT-5.60.1417$64.2780.6M214k10h 2m3893
38Muse Spark 1.20.1353$32.24170.7M487k10h 37m5515
39GLM-5.30.1261$28.9386.9M409k13h 16m3282
40Muse Spark 1.20.0816$32.40192.3M534k7h 46m5704
41Gemini 3.7 Flash0.0417$2.9916.0M112k3h 8m1656
42Gemini 3.7 Flash0.0412$3.0519.8M150k3h 10m19410
43Gemini 3.7 Flash0.0403$4.5617.9M128k4h 20m1458
44Muse Spark 1.20.0387$183.05621.0M1.3M19h 7m14977
45Muse Spark 1.20.0356$37.95196.7M1.2M19h 2m6693
46Inkling0.0214$5.305.4M31k3h 11m1842
47Inkling0.0209$6.766.9M44k2h 24m1932
48Inkling0.0205$8.128.4M30k3h 20m1851
49Inkling0.0203$5.445.6M27k1h 39m1213
50Inkling0.0179$5.866.0M36k2h 37m1662

Flight-Sim Renderer in OpenGL

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.4716$208.91376.5M866k17h 50m101710
2Claude Fable 5.10.3532$168.41314.0M539k12h 53m6084
3Claude Fable 5.10.3481$106.70193.6M429k9h 8m45413
4Claude Fable 5.10.3402$125.75229.9M473k10h 25m4826
5GPT-5.60.2171$270.86274.0M362k12h 30m8064
6Claude Fable 5.10.1985$113.40205.2M463k5h 28m4303
7GPT-5.60.1552$336.67340.6M347k11h 28m90614
8GPT-5.60.1222$253.61257.1M305k8h 28m7603
9GPT-5.60.1119$309.74312.3M342k9h 22m85712
10GPT-5.60.1032$420.32406.2M354k11h 35m88013
11Grok 4.60.0526$317.62274.4M1.8M19h 55m1078110
12Grok 4.60.0382$286.38260.8M1.4M19h 57m984103
13Grok 4.60.0273$488.84437.7M2.0M19h 56m1536307
14Grok 4.60.0266$454.46403.4M2.0M19h 58m143575
15Kimi K30.0249$126.07350.4M1.1M19h 56m9367
16Kimi K30.0214$168.60476.4M1.3M20h 1m10795
17GLM-5.30.0166$91.30316.9M852k19h 35m8917
18Grok 4.60.0147$328.36283.8M1.8M19h 57m108457
19Gemini 3.7 Flash0.0116$11.2868.7M237k4h 7m3042
20Kimi K30.0104$167.00460.9M1.4M20h 1m10391
21Kimi K30.0079$184.18516.4M1.3M19h 37m10942
22DeepSeek V4 Flash Vision Exp0.0069$10.96663.9M1.1M19h 2m15272
23DeepSeek V4 Flash Vision Exp0.0065$10.60633.4M1.1M15h 56m15168
24Gemini 3.7 Flash0.0064$35.75195.3M360k8h 28m5551
25GLM-5.30.0054$111.06362.5M755k19h 29m89310
26Gemini 3.7 Flash0.0054$6.5251.3M224k2h 40m2742
27GLM-5.30.0038$97.66297.0M573k19h 4m6335
28DeepSeek V4 Flash Vision Exp0.0028$12.40756.6M1.2M19h 8m175639
29DeepSeek V4 Flash Vision Exp0.0027$6.77415.2M638k18h 28m8377
30Gemini 3.7 Flash0.0020$17.05105.9M300k5h 32m3811
31Qwen3.8-Max0.0018$183.02558.1M1.7M19h 56m13751
32Qwen3.8-Max0.0016$153.43448.8M1.7M19h 59m12021
33Qwen3.8-Max0.0010$149.46419.0M1.2M19h 25m9932
34Gemini 3.7 Flash0.0009$11.2877.4M230k5h 1m3693
35GLM-5.30.0006$94.18300.9M621k18h 57m59210
36Qwen3.8-Max0.0001$169.64519.3M1.6M20h 0m12391
37Muse Spark 1.20.0000$7.7740.8M287k2h 9m1825
38Inkling0.0000$6.727.0M25k22m1382
39Qwen3.8-Max0.0000$189.91559.5M1.9M19h 51m14643
40Kimi K30.0000$182.81498.0M1.4M20h 1m10632
41Muse Spark 1.20.0000$8.0738.9M412k3h 7m2024
42DeepSeek V4 Flash Vision Exp0.0000$11.01666.7M1.1M19h 11m146711
43Muse Spark 1.20.0000$6.3024.2M135k1h 37m1594
44GLM-5.30.0000$103.09375.1M554k19h 46m79012
45Inkling0.0000$6.596.8M31k2h 10m1462
46Muse Spark 1.20.0000$2.6412.4M42k2h 25m902
47Inkling0.0000$6.576.8M31k35m1561
48Inkling0.0000$4.244.4M17k41m1091
49Inkling0.0000$8.879.2M43k1h 49m1702
50Muse Spark 1.20.0000$1.728.1M39k1h 17m881

† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.

Optimizer Design

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1GPT-5.60.3487$65.5380.8M133k10h 22m4334
2GPT-5.60.3194$196.66203.2M202k11h 49m67823
3Kimi K30.3057$15.0637.2M181k19h 11m2754
4GPT-5.60.2846$55.2972.7M133k15h 42m4235
5GPT-5.60.2809$71.6593.2M177k14h 22m52321
6Grok 4.60.2557$99.1498.0M624k19h 20m42418
7GLM-5.30.2349$19.9835.0M224k17h 41m1923
8Claude Fable 5.10.2272$15.4522.4M175k11h 13m1645
9GPT-5.60.2234$31.2952.9M100k13h 17m3732
10GLM-5.30.2192$17.3736.9M195k19h 6m2412
11GLM-5.30.2082$16.4826.8M198k19h 23m1843
12Grok 4.60.1957$53.9066.7M503k19h 24m39524
13Claude Fable 5.10.1750$17.1526.3M165k12h 46m1826
14Grok 4.60.1570$59.5357.5M295k19h 26m25811
15Kimi K30.1475$41.51100.0M508k20h 1m3544
16Claude Fable 5.10.1459$14.9022.4M152k15h 9m1714
17Kimi K30.1370$16.0538.0M228k19h 25m2224
18Grok 4.60.1368$54.4949.3M462k19h 16m24010
19Claude Fable 5.10.1343$11.0815.3M141k13h 28m1343
20Gemini 3.7 Flash0.1194$22.13241.0M134k10h 36m8342
21DeepSeek V4 Flash Vision Exp0.1123$1.1252.3M255k18h 23m2562
22Kimi K30.0990$22.5849.0M297k17h 53m2416
23GLM-5.30.0984$17.0155.5M261k19h 4m2775
24GLM-5.30.0965$13.4723.0M159k18h 59m1802
25Grok 4.60.0839$80.4575.7M333k19h 42m2793
26Qwen3.8-Max0.0827$7.6817.4M516k19h 49m2263
27DeepSeek V4 Flash Vision Exp0.0770$1.2263.5M222k19h 1m3342
28Qwen3.8-Max0.0741$7.6118.8M448k19h 35m2413
29Qwen3.8-Max0.0655$6.8715.9M443k19h 59m2033
30Gemini 3.7 Flash0.0537$10.3785.1M122k13h 22m4352
31Gemini 3.7 Flash0.0405$25.34217.9M182k10h 29m7442
32DeepSeek V4 Flash Vision Exp0.0348$1.4475.8M242k19h 19m3492
33Claude Fable 5.10.0226$15.7421.7M202k14h 5m1414
34Qwen3.8-Max0.0143$7.3218.3M421k19h 31m2381
35Muse Spark 1.20.0000$39.2585.0M351k13h 51m2757
36Muse Spark 1.20.0000$46.70116.8M475k9h 50m3648
37DeepSeek V4 Flash Vision Exp0.0000$1.5179.4M262k19h 14m4062
38Muse Spark 1.20.0000$55.59163.2M672k19h 37m5671
39Gemini 3.7 Flash0.0000$28.88259.5M185k18h 35m9081
40Gemini 3.7 Flash0.0000$12.4147.0M144k13h 48m2972
41DeepSeek V4 Flash Vision Exp0.0000$1.3365.8M275k19h 51m3131
42Muse Spark 1.20.0000$27.0448.5M345k6h 22m2563
43Muse Spark 1.20.0000$16.2119.8M148k19h 57m1881
44Inkling0.0000$1.081.1M14k1h 4m581
45Kimi K30.0000$19.0542.1M297k14h 45m2193
46Inkling0.0000$1.531.5M16k1h 13m732
47Qwen3.8-Max0.0000$12.4233.0M636k19h 59m3031
48Inkling0.0000$3.843.9M25k5h 18m1241
49Inkling0.0000$2.542.6M28k2h 1m881
50Inkling0.0000$1.631.6M27k1h 46m731

PostgreSQL 18 on SQLite

task page →49 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1GPT-5.60.5275$209.42198.9M277k4h 5m58912
2Claude Fable 5.10.4011$588.091056.3M2.5M17h 46m251513
3Claude Fable 5.10.3901$721.011319.3M2.5M17h 44m287214
4GPT-5.60.3901$654.39640.1M687k19h 58m147228
5Claude Fable 5.10.3516$384.99680.6M1.8M18h 59m16889
6Claude Fable 5.10.3352$475.86850.2M2.1M18h 4m18947
7Gemini 3.7 Flash0.2582$124.971333.8M2.5M11h 51m301437
8Gemini 3.7 Flash0.2418$144.251592.6M2.6M19h 46m345039
9Claude Fable 5.10.2308$392.90697.7M1.8M19h 16m16794
10Gemini 3.7 Flash0.2253$193.892112.9M3.0M19h 55m468016
11Gemini 3.7 Flash0.1758$159.991689.8M2.8M19h 49m375113
12Grok 4.60.1593$509.89496.5M1.6M19h 56m169670
13GLM-5.30.1593$191.86711.5M862k19h 53m174617
14Grok 4.60.1538$461.37465.7M1.9M17h 59m184249
15DeepSeek V4 Flash Vision Exp0.1484$22.501379.3M2.0M19h 53m31461
16GLM-5.30.1484$202.20744.7M1.0M19h 42m17897
17GLM-5.30.1374$201.89706.7M843k19h 32m15859
18GLM-5.30.1374$155.77570.5M772k19h 52m13842
19Qwen3.8-Max0.1374$186.86700.0M1.6M19h 30m18222
20Qwen3.8-Max0.1319$178.70667.7M1.7M19h 47m15494
21DeepSeek V4 Flash Vision Exp0.1264$22.611383.3M2.0M19h 51m32723
22DeepSeek V4 Flash Vision Exp0.1044$19.661214.6M1.7M19h 25m26889
23Grok 4.60.0989$257.56255.7M1.5M8h 44m95217
24GLM-5.30.0989$219.76822.5M966k19h 50m190311
25Qwen3.8-Max0.0769$170.86638.4M1.5M19h 53m15872
26Grok 4.60.0714$398.74383.4M2.7M19h 51m145734
27Kimi K30.0714$227.56673.8M1.2M20h 0m15281
28DeepSeek V4 Flash Vision Exp0.0659$23.231428.3M2.0M19h 44m33481
29Qwen3.8-Max0.0659$174.14653.5M1.5M19h 26m15361
30Kimi K30.0659$265.37772.2M1.4M20h 1m14271
31Kimi K30.0604$238.20702.8M1.4M20h 1m15131
32Qwen3.8-Max0.0549$158.21583.9M1.7M19h 23m14053
33Kimi K30.0495$197.52561.9M1.3M20h 0m11681
34Grok 4.60.0495$148.44141.0M985k6h 17m55910
35Gemini 3.7 Flash0.0495$17.96182.1M620k1h 34m4022
36DeepSeek V4 Flash Vision Exp0.0165$8.18489.5M856k20h 0m11942
37Muse Spark 1.20.0055$37.82217.7M976k3h 48m49610
38Muse Spark 1.20.0055$40.72146.5M561k3h 14m4204
39GPT-5.60.0055$293.05285.8M311k4h 20m6716
40Muse Spark 1.20.0000$32.40176.5M473k2h 53m4363
41Inkling0.0000$22.3423.3M56k1h 53m3403
42Muse Spark 1.20.0000$14.0883.6M240k1h 8m3142
43GPT-5.60.0000$137.59141.5M243k3h 27m4802
44GPT-5.60.0000$142.29142.1M259k4h 34m46211
45Inkling0.0000$19.0619.8M53k1h 41m3022
46Inkling0.0000$19.0419.9M46k23m2521
47Inkling0.0000$16.5617.2M48k32m2142
48Inkling0.0000$16.6217.3M42k34m2871
49Muse Spark 1.20.0000$33.19162.4M603k2h 4m4581

† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.

Quantum ESPRESSO pw.x in Rust

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.9431$100.68158.8M876k5h 35m40811
2Claude Fable 5.10.9350$63.8099.1M584k3h 4m2188
3Claude Fable 5.10.9268$63.9599.7M578k3h 23m2225
4Claude Fable 5.10.9187$72.59116.9M580k4h 7m2455
5Claude Fable 5.10.9106$77.90124.4M643k3h 55m2564
6GPT-5.60.5203$304.21296.3M260k4h 50m6069
7GPT-5.60.4634$57.4264.9M189k2h 47m27913
8GPT-5.60.4553$120.99116.9M214k5h 5m4048
9GPT-5.60.4065$104.97107.5M204k3h 16m39110
10Grok 4.60.3984$196.95181.0M1.6M13h 51m68114
11Grok 4.60.3008$282.67273.1M1.6M13h 49m97640
12Grok 4.60.2683$1211.081191.6M2.2M19h 59m3559243
13Kimi K30.2602$75.00147.3M1.2M19h 51m4281
14GLM-5.30.1707$145.94507.1M1.3M19h 30m108213
15Gemini 3.7 Flash0.1545$46.71314.3M747k10h 28m7402
16GLM-5.30.1382$132.96476.8M1.3M19h 42m98412
17GLM-5.30.1301$108.12382.4M1.3M19h 32m8382
18Grok 4.60.1057$123.16111.5M1.7M10h 28m42913
19Grok 4.60.0976$96.1488.3M1.1M7h 19m36012
20Gemini 3.7 Flash0.0976$71.79423.8M777k17h 50m9483
21Gemini 3.7 Flash0.0650$36.49288.8M712k6h 10m7252
22GLM-5.30.0650$123.53403.7M1.6M19h 19m95711
23Kimi K30.0488$160.13415.5M1.7M20h 0m9991
24Muse Spark 1.20.0244$48.27287.1M1.0M3h 0m5981
25Kimi K30.0244$100.42211.3M1.8M20h 0m5661
26Kimi K30.0081$131.87355.0M1.3M19h 56m7856
27Qwen3.8-Max0.0000$48.53146.5M1.8M19h 49m4311
28DeepSeek V4 Flash Vision Exp0.0000$7.62422.1M1.1M12h 5m9126
29Gemini 3.7 Flash0.0000$51.00393.0M804k11h 3m10024
30Kimi K30.0000$72.76158.2M1.2M20h 1m4181
31Muse Spark 1.20.0000$22.0858.0M685k1h 30m1943
32Qwen3.8-Max0.0000$54.35159.6M2.3M19h 59m4131
33Qwen3.8-Max0.0000$33.3786.6M1.8M19h 52m3012
34Gemini 3.7 Flash0.0000$24.96238.1M690k3h 15m5392
35Inkling0.0000$18.6819.4M66k28m1862
36DeepSeek V4 Flash Vision Exp0.0000$6.14336.4M959k16h 29m85310
37Inkling0.0000$22.9424.0M46k17m2741
38DeepSeek V4 Flash Vision Exp0.0000$4.17227.6M660k6h 35m4473
39GPT-5.60.0000$210.32203.5M250k6h 57m5776
40Inkling0.0000$8.198.5M35k16m1421
41Muse Spark 1.20.0000$8.0039.5M393k1h 21m1662
42Inkling0.0000$9.259.5M49k17m1462
43DeepSeek V4 Flash Vision Exp0.0000$6.64359.4M1.1M12h 22m8825
44GLM-5.30.0000$103.68348.9M1.2M19h 23m79514
45DeepSeek V4 Flash Vision Exp0.0000$6.97389.0M1.0M14h 29m8875
46Muse Spark 1.20.0000$5.4530.5M131k29m1603
47Qwen3.8-Max0.0000$35.4395.7M1.8M20h 1m3381
48Inkling0.0000$9.329.6M44k15m1382
49Muse Spark 1.20.0000$10.3054.9M392k1h 20m2484
50Qwen3.8-Max0.0000$11.6213.4M1.3M19h 53m1121

Qubit Routing

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.8857$37.5960.9M295k9h 18m2396
2Claude Fable 5.10.8752$40.6567.2M290k7h 39m2749
3Claude Fable 5.10.8680$40.9467.5M295k10h 23m2799
4Claude Fable 5.10.8599$41.2367.7M303k9h 29m2569
5Claude Fable 5.10.8541$59.53101.1M368k11h 5m3537
6GPT-5.60.7946$58.7777.9M203k6h 45m4238
7GPT-5.60.7781$75.1186.9M202k10h 41m41011
8Gemini 3.7 Flash0.7591$2.339.2M139k1h 43m992
9GLM-5.30.7535$23.4664.7M277k19h 3m2676
10Kimi K30.7526$40.44107.3M401k19h 37m38717
11Grok 4.60.7350$123.13116.9M502k9h 53m41428
12GLM-5.30.7343$45.4890.2M303k16h 27m3654
13GPT-5.60.7314$114.07131.6M228k8h 56m56914
14GPT-5.60.7190$154.11169.8M283k6h 40m68813
15Kimi K30.7151$55.15146.5M497k20h 0m48414
16GPT-5.60.7125$69.3392.1M185k8h 7m4755
17Kimi K30.6987$79.07224.2M641k20h 1m55612
18GLM-5.30.6907$44.40135.6M383k15h 8m39212
19Grok 4.60.6883$68.0360.3M544k3h 28m2496
20Qwen3.8-Max0.6875$16.9452.6M563k18h 44m3402
21Grok 4.60.6689$82.9075.9M477k3h 6m28317
22Gemini 3.7 Flash0.6672$3.0415.3M126k1h 37m1401
23Grok 4.60.6619$46.4748.9M429k4h 33m2747
24Gemini 3.7 Flash0.6446$3.4416.9M214k2h 28m1512
25GLM-5.30.6370$18.5955.0M231k19h 13m2873
26Kimi K30.6270$30.7976.7M394k12h 20m31820
27Kimi K30.6254$97.66258.3M819k19h 58m57617
28DeepSeek V4 Flash Vision Exp0.6244$2.53133.2M457k10h 50m4027
29Muse Spark 1.20.6114$9.2347.6M407k3h 16m2624
30Gemini 3.7 Flash0.6074$16.2672.9M429k6h 43m2805
31GLM-5.30.6071$32.20106.0M372k12h 23m3755
32Muse Spark 1.20.5947$9.3645.3M339k3h 54m2442
33Grok 4.60.5924$84.0182.0M496k4h 20m33610
34Muse Spark 1.20.5914$9.2141.8M337k3h 53m2302
35Gemini 3.7 Flash0.5912$5.6344.5M188k5h 26m2701
36Muse Spark 1.20.5787$10.6452.8M389k2h 30m2728
37Qwen3.8-Max0.5728$25.9085.8M648k18h 2m4426
38Muse Spark 1.20.5545$7.7042.7M226k2h 46m2575
39DeepSeek V4 Flash Vision Exp0.5485$2.27120.4M400k9h 24m4053
40DeepSeek V4 Flash Vision Exp0.4883$3.26177.3M519k9h 8m4267
41DeepSeek V4 Flash Vision Exp0.4768$2.33127.1M362k12h 24m4074
42Qwen3.8-Max0.4459$25.1978.5M819k19h 40m3742
43DeepSeek V4 Flash Vision Exp0.4348$2.35127.0M382k9h 0m3957
44Inkling0.4268$4.274.3M39k1h 28m1012
45Qwen3.8-Max0.4074$19.9763.7M596k17h 56m3902
46Qwen3.8-Max0.2765$43.05138.5M1.3M20h 1m5121
47Inkling0.2712$3.613.7M31k1h 49m902
48Inkling0.2062$7.667.9M44k1h 46m1582
49Inkling0.1413$4.714.8M39k39m982
50Inkling0.0057$3.813.8M60k1h 38m951

Reconnaissance Blind Chess Recovery

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.9536$56.7098.3M311k9h 23m3147
2Claude Fable 5.10.9536$46.0878.9M273k10h 59m2769
3Claude Fable 5.10.9429$31.3851.1M239k11h 53m2336
4GPT-5.60.8893$28.4245.1M149k6h 33m2917
5GPT-5.60.8607$35.9957.3M138k6h 13m33613
6Claude Fable 5.10.8357$41.8471.9M245k9h 32m25211
7GPT-5.60.8071$20.6232.5M99k10h 50m19610
8GPT-5.60.8000$20.5330.5M123k2h 52m1973
9Claude Fable 5.10.7929$38.2264.0M257k7h 37m2454
10Kimi K30.7571$29.2377.1M277k14h 20m3142
11GPT-5.60.7357$15.5121.1M106k11h 38m1689
12Grok 4.60.6036$87.2970.2M604k8h 33m2599
13Grok 4.60.5857$104.5199.3M819k14h 17m45324
14GLM-5.30.5357$48.03138.0M410k14h 33m3889
15Kimi K30.5071$31.5182.4M325k14h 46m3114
16Kimi K30.4286$27.4364.3M286k14h 50m25611
17Gemini 3.7 Flash0.4250$12.0951.0M293k4h 11m2663
18GLM-5.30.4214$29.5279.9M302k14h 26m3056
19Grok 4.60.4071$107.3799.7M852k13h 36m44317
20Muse Spark 1.20.4036$5.4124.7M331k4h 44m1502
21Muse Spark 1.20.4036$6.7330.6M391k4h 23m1324
22Muse Spark 1.20.3750$11.1857.5M431k6h 54m2546
23GLM-5.30.3607$36.0573.5M339k13h 30m2594
24Muse Spark 1.20.3429$9.0140.4M451k7h 47m1974
25Kimi K30.3143$28.9169.9M341k12h 50m2532
26Gemini 3.7 Flash0.3000$23.11214.3M578k5h 22m5241
27GLM-5.30.2893$25.2683.8M284k14h 38m3394
28Grok 4.60.2786$78.3465.1M979k10h 29m3018
29GLM-5.30.2571$27.0269.8M315k14h 43m2536
30Gemini 3.7 Flash0.2393$20.54111.4M562k4h 43m3271
31Muse Spark 1.20.2071$9.6939.5M748k4h 22m1387
32Inkling0.2000$1.231.2M15k1h 37m461
33DeepSeek V4 Flash Vision Exp0.1786$2.02107.0M337k14h 34m3282
34Kimi K30.1429$43.61116.1M397k14h 40m3952
35Qwen3.8-Max0.1357$20.2059.6M794k13h 51m3453
36DeepSeek V4 Flash Vision Exp0.1214$2.99164.7M448k8h 44m4235
37Grok 4.60.0929$172.71159.1M1.1M14h 32m60914
38Inkling0.0786$1.371.4M14k43m621
39Qwen3.8-Max0.0750$16.7049.8M638k13h 27m3353
40Qwen3.8-Max0.0536$12.9235.1M623k13h 47m2383
41Gemini 3.7 Flash0.0536$30.51186.7M1.3M5h 45m5132
42Qwen3.8-Max0.0536$22.7970.5M780k13h 7m4153
43Gemini 3.7 Flash0.0429$30.88192.0M568k8h 45m5062
44Inkling0.0393$5.275.4M45k1h 49m1272
45DeepSeek V4 Flash Vision Exp0.0357$2.71145.6M448k14h 17m3777
46DeepSeek V4 Flash Vision Exp0.0143$2.49130.6M440k11h 4m40010
47Inkling0.0143$2.102.1M21k28m831
48Qwen3.8-Max0.0071$26.0582.8M805k13h 22m4363
49DeepSeek V4 Flash Vision Exp0.0036$2.74145.9M463k9h 28m3726
50Inkling0.0000$3.914.0M36k2h 12m982

Fitness-Recap Video in Remotion

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1GPT-5.60.2152$950.51938.4M1.0M19h 3m234345
2Claude Fable 5.10.2055$157.69293.5M491k6h 51m6658
3Claude Fable 5.10.2018$157.38293.1M476k6h 7m6447
4GPT-5.60.1930$656.93636.3M650k19h 12m147811
5Claude Fable 5.10.1927$120.12223.7M414k5h 14m5877
6GPT-5.60.1913$298.28288.7M261k7h 23m65112
7GPT-5.60.1895$283.99275.5M274k5h 43m6876
8GLM-5.30.1894$78.50212.7M442k18h 20m5306
9GLM-5.30.1828$113.14407.8M483k19h 2m8049
10Kimi K30.1767$191.75571.0M982k19h 53m13563
11Claude Fable 5.10.1730$560.601048.2M1.5M18h 2m227426
12Kimi K30.1723$171.83504.1M1.1M19h 44m12348
13Kimi K30.1716$266.42802.2M1.3M20h 1m16181
14DeepSeek V4 Flash Vision Exp0.1621$10.73658.6M994k19h 13m165118
15Qwen3.8-Max0.1474$77.77271.1M994k13h 7m8004
16GLM-5.30.1426$118.44380.0M784k18h 55m98618
17GPT-5.60.1374$190.12189.9M230k5h 21m5892
18GLM-5.30.1053$110.52388.9M908k15h 36m9619
19GLM-5.30.0837$90.93307.2M485k16h 10m6368
20DeepSeek V4 Flash Vision Exp0.0738$6.53405.2M565k19h 26m99410
21Qwen3.8-Max0.0733$163.48458.4M1.7M19h 54m12491
22Claude Fable 5.10.0694$335.29627.1M960k15h 45m145619
23Kimi K30.0657$243.01727.7M1.1M20h 1m16331
24Kimi K30.0536$202.63587.9M1.0M19h 57m147912
25DeepSeek V4 Flash Vision Exp0.0483$6.64411.0M525k16h 0m98321
26Grok 4.60.0338$1762.941204.0M2.8M20h 0m4175289
27DeepSeek V4 Flash Vision Exp0.0324$12.46767.8M1.1M18h 4m178924
28Grok 4.60.0279$646.10557.5M1.5M19h 48m2043149
29Grok 4.60.0245$555.19490.3M1.5M19h 56m202090
30Grok 4.60.0242$1149.83792.5M1.8M19h 58m2836250
31Grok 4.60.0216$746.94531.5M1.4M19h 55m1941163
32Gemini 3.7 Flash0.0129$13.97114.7M211k4h 23m3822
33Qwen3.8-Max0.0127$196.38482.4M1.8M19h 51m10481
34DeepSeek V4 Flash Vision Exp0.0121$4.01242.4M395k12h 19m77716
35Gemini 3.7 Flash0.0091$2.9421.2M86k1h 20m1692
36Gemini 3.7 Flash0.0089$10.4467.7M164k2h 48m3284
37Gemini 3.7 Flash0.0080$7.6861.7M148k2h 7m3001
38Muse Spark 1.20.0079$12.4662.2M180k3h 28m2703
39Qwen3.8-Max0.0062$52.83157.6M675k9h 28m5784
40Muse Spark 1.20.0060$91.70184.1M1.3M8h 42m5093
41Gemini 3.7 Flash0.0060$4.6730.1M112k2h 7m2221
42Muse Spark 1.20.0032$14.7129.2M224k3h 30m1832
43Muse Spark 1.20.0031$37.8575.3M303k4h 52m2854
44Inkling0.0006$14.0514.6M38k1h 0m1831
45Inkling0.0001$7.968.3M26k31m1412
46Inkling0.0000$13.7214.2M49k59m2051
47Inkling0.0000$10.6511.1M29k1h 18m1722
48Inkling0.0000$11.1511.6M44k40m1381
49Qwen3.8-Max0.0000$11.9131.2M238k4h 2m2922
50Muse Spark 1.20.0000$12.6638.1M209k4h 41m1901

† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.

Snooker Prediction

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.5126$87.17154.8M404k6h 52m40812
2Claude Fable 5.10.4984$96.24173.0M403k7h 34m48621
3GPT-5.60.4857$119.18111.7M242k2h 5m29911
4GPT-5.60.4839$170.16165.0M275k4h 12m4329
5Claude Fable 5.10.4822$86.06152.3M409k8h 44m40223
6Claude Fable 5.10.4767$88.31156.7M413k3h 39m39710
7GPT-5.60.4428$157.87150.8M208k2h 34m3549
8GPT-5.60.4295$132.32126.4M240k4h 1m3696
9Grok 4.60.3789$466.11391.2M2.0M19h 43m148873
10Claude Fable 5.10.3738$86.02153.2M388k5h 28m4249
11Grok 4.60.3048$106.1286.0M725k5h 58m3785
12GPT-5.60.2890$179.27154.6M213k9h 23m3575
13GLM-5.30.1976$71.12251.5M765k18h 56m65415
14Grok 4.60.1779$108.4587.9M851k6h 23m34415
15Grok 4.60.1175$125.27104.6M768k11h 27m4469
16GLM-5.30.1013$57.78201.7M793k14h 16m53113
17Qwen3.8-Max0.0882$56.82135.8M1.2M18h 54m5061
18GLM-5.30.0590$53.37163.6M430k17h 18m3736
19Inkling0.0335$2.612.6M26k1h 16m722
20GLM-5.30.0311$89.97246.3M519k19h 15m49513
21Grok 4.60.0252$77.1660.3M816k4h 29m27113
22Inkling0.0203$2.912.9M46k1h 53m872
23Kimi K30.0000$126.53346.0M738k19h 31m7788
24Qwen3.8-Max0.0000$80.65199.9M2.1M18h 39m6731
25Gemini 3.7 Flash0.0000$7.8266.5M374k1h 5m2791
26Gemini 3.7 Flash0.0000$8.3359.1M343k2h 40m2823
27Inkling0.0000$3.243.2M40k1h 45m1031
28Gemini 3.7 Flash0.0000$5.6140.2M288k59m2242
29Muse Spark 1.20.0000$26.92100.5M603k1h 48m3775
30DeepSeek V4 Flash Vision Exp0.0000$4.83277.6M638k5h 11m64413
31Muse Spark 1.20.0000$12.0849.7M275k1h 6m2508
32DeepSeek V4 Flash Vision Exp0.0000$3.23179.6M479k19h 58m4661
33DeepSeek V4 Flash Vision Exp0.0000$3.55189.8M629k6h 0m4853
34Qwen3.8-Max0.0000$115.97289.9M2.1M19h 35m8111
35Qwen3.8-Max0.0000$102.79307.3M1.7M19h 10m7942
36Gemini 3.7 Flash0.0000$6.6448.6M323k1h 55m2471
37Kimi K30.0000$140.38372.9M1.4M19h 51m92526
38Inkling0.0000$2.842.7M69k45m822
39Muse Spark 1.20.0000$16.1178.5M517k1h 19m3094
40Qwen3.8-Max0.0000$61.40180.5M1.6M20h 1m5511
41Inkling0.0000$1.801.8M26k14m792
42Muse Spark 1.20.0000$13.7075.9M319k1h 35m3053
43Kimi K30.0000$122.14346.5M798k19h 47m7938
44Gemini 3.7 Flash0.0000$7.4056.9M413k1h 21m2662
45DeepSeek V4 Flash Vision Exp0.0000$4.28248.4M549k14h 16m5938
46Kimi K30.0000$117.31316.7M1.0M19h 47m8066
47Kimi K30.0000$105.21273.3M660k20h 0m5483
48GLM-5.30.0000$94.15290.7M465k18h 52m62711
49DeepSeek V4 Flash Vision Exp0.0000$5.35314.7M642k14h 37m7093
50Muse Spark 1.20.0000$16.5887.1M508k2h 14m3216

SPICE Circuit Simulator in Rust

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.8586$166.41276.3M1.2M9h 57m63217
2Claude Fable 5.10.7677$93.05145.4M837k3h 54m4174
3Claude Fable 5.10.7071$96.79153.4M827k4h 2m4138
4Claude Fable 5.10.6970$69.08107.8M623k2h 37m2272
5Claude Fable 5.10.6667$96.05148.5M898k4h 16m4066
6Gemini 3.7 Flash0.6263$99.311037.7M2.0M6h 16m25774
7Grok 4.60.5657$717.25714.4M2.7M18h 22m2483266
8GLM-5.30.5354$171.96618.6M1.7M18h 15m14414
9Gemini 3.7 Flash0.5253$121.421230.5M2.0M12h 5m29493
10Grok 4.60.5253$835.65836.5M3.1M20h 0m3128205
11GLM-5.30.4949$235.96867.9M1.6M17h 11m190518
12Grok 4.60.4848$586.10582.0M3.0M17h 34m232289
13Gemini 3.7 Flash0.4747$76.85790.2M1.6M4h 46m20172
14Grok 4.60.4646$1191.201187.2M2.5M19h 58m387996
15Grok 4.60.4545$165.93154.1M860k5h 37m56337
16Gemini 3.7 Flash0.4444$80.51766.1M1.3M6h 23m17937
17GPT-5.60.4444$366.07350.2M455k6h 25m79722
18GLM-5.30.4444$158.88578.0M1.3M10h 53m11554
19GLM-5.30.4141$190.24693.2M1.5M19h 41m170530
20GPT-5.60.4141$307.32296.1M386k4h 30m78315
21GPT-5.60.3939$538.95528.2M617k6h 30m128021
22Gemini 3.7 Flash0.3737$126.281246.3M2.0M12h 6m29212
23Qwen3.8-Max0.3737$145.98524.1M2.1M19h 26m13381
24Qwen3.8-Max0.3434$127.04463.9M1.6M19h 13m9874
25GPT-5.60.3333$187.27186.7M262k2h 15m5158
26GLM-5.30.3131$152.26545.3M1.3M16h 57m116423
27Qwen3.8-Max0.3030$96.60342.7M1.6M18h 43m7602
28Qwen3.8-Max0.3030$110.14386.5M2.0M19h 56m9471
29Qwen3.8-Max0.2727$126.58455.7M1.9M19h 26m9517
30GPT-5.60.2626$226.84221.2M327k3h 7m59213
31Muse Spark 1.20.1313$32.05183.6M731k2h 8m4651
32Muse Spark 1.20.0707$10.8153.3M351k1h 35m2232
33Muse Spark 1.20.0707$32.82175.7M815k3h 21m4611
34Muse Spark 1.20.0505$8.7748.8M242k1h 16m1962
35Muse Spark 1.20.0404$5.3427.9M188k26m1432
36DeepSeek V4 Flash Vision Exp0.0000$11.65704.6M1.2M10h 40m154915
37Kimi K30.0000$179.57474.0M1.9M19h 42m12159
38Kimi K30.0000$215.15588.2M1.9M19h 56m14053
39DeepSeek V4 Flash Vision Exp0.0000$19.651192.1M2.0M18h 3m239810
40Kimi K30.0000$199.08565.4M1.4M19h 58m12711
41Kimi K30.0000$204.94570.3M1.7M19h 34m13993
42Kimi K30.0000$193.03529.1M1.8M19h 58m12631
43DeepSeek V4 Flash Vision Exp0.0000$20.481199.6M2.4M11h 32m25639
44Inkling0.0000$4.424.6M23k45m831
45DeepSeek V4 Flash Vision Exp0.0000$21.511287.0M2.2M19h 10m290413
46Inkling0.0000$19.3020.1M40k1h 20m1702
47Inkling0.0000$1.041.1M10k27m522
48DeepSeek V4 Flash Vision Exp0.0000$19.681182.5M2.0M19h 8m261723
49Inkling0.0000$5.075.3M18k1h 13m1102
50Inkling0.0000$5.215.3M32k15m1242

Synthetic Music Diarization

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.5457$12.1118.5M118k17h 49m1568
2Claude Fable 5.10.5346$11.4117.0M120k17h 44m1587
3GPT-5.60.5247$44.1953.3M154k7h 34m2673
4Claude Fable 5.10.5104$17.6028.7M134k17h 45m23011
5GPT-5.60.5065$80.6578.5M178k15h 57m33911
6GPT-5.60.4952$59.6269.7M147k15h 37m3593
7Claude Fable 5.10.4905$10.2915.1M114k17h 40m1407
8Claude Fable 5.10.4889$10.8615.6M125k15h 34m1386
9GLM-5.30.4816$35.34125.4M361k17h 30m46211
10GLM-5.30.4802$31.6279.0M303k18h 9m3611
11Gemini 3.7 Flash0.4776$32.86129.1M265k12h 34m4382
12GPT-5.60.4737$78.7876.5M133k14h 47m2864
13GPT-5.60.4727$68.5982.8M153k14h 29m4257
14Gemini 3.7 Flash0.4664$11.2742.4M155k7h 29m2542
15Grok 4.60.4661$133.75133.0M655k19h 52m57842
16Qwen3.8-Max0.4605$29.62103.3M529k18h 43m50911
17GLM-5.30.4602$19.7741.7M196k19h 9m2534
18Qwen3.8-Max0.4425$26.6693.3M455k18h 13m4788
19Gemini 3.7 Flash0.4422$14.36133.2M250k5h 49m3892
20DeepSeek V4 Flash Vision Exp0.4412$1.7293.1M282k19h 4m4302
21Grok 4.60.4323$188.19192.5M1.1M19h 45m87156
22Kimi K30.4318$13.3633.2M179k18h 52m2445
23Qwen3.8-Max0.4301$34.00119.2M572k19h 17m50810
24GLM-5.30.4285$28.0377.4M258k18h 22m3669
25GLM-5.30.4265$27.2778.5M240k19h 37m3746
26Gemini 3.7 Flash0.4264$42.51256.4M394k13h 10m6819
27DeepSeek V4 Flash Vision Exp0.4247$2.96173.2M352k17h 55m6174
28Kimi K30.4157$22.8160.7M202k19h 10m3477
29Grok 4.60.4120$228.84232.3M993k19h 44m91863
30DeepSeek V4 Flash Vision Exp0.4065$1.7395.5M262k17h 34m4485
31Qwen3.8-Max0.4043$33.94117.1M655k18h 57m5015
32Kimi K30.4034$13.3431.5M191k18h 19m2474
33Gemini 3.7 Flash0.4000$15.6783.4M201k4h 25m3221
34Grok 4.60.3985$299.53304.5M1.6M19h 27m128154
35Qwen3.8-Max0.3951$17.0256.1M424k19h 10m3914
36Kimi K30.3852$16.4839.9M237k19h 5m2342
37Muse Spark 1.20.3829$88.55356.6M541k7h 12m7543
38DeepSeek V4 Flash Vision Exp0.3803$1.8599.9M306k18h 37m4553
39Muse Spark 1.20.3711$10.8742.0M256k3h 30m2274
40Kimi K30.3635$16.0039.1M227k18h 59m2346
41Muse Spark 1.20.3611$12.5032.3M167k4h 4m2135
42Muse Spark 1.20.3544$13.5350.0M335k4h 26m2484
43Muse Spark 1.20.3434$13.4047.8M217k5h 21m2415
44DeepSeek V4 Flash Vision Exp0.2979$2.02115.0M266k19h 32m5084
45Inkling0.2089$6.226.4M44k35m1312
46Inkling0.0477$6.406.6M38k28m1231
47Inkling0.0074$12.0812.5M45k59m1601
48Inkling0.0053$5.675.8M48k1h 31m992
49Inkling0.0012$9.519.9M36k38m1701
50Grok 4.60.0000$501.54509.4M1.9M18h 17m193796

Vision-only TORCS Racing Bot

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.3604$65.32119.8M246k14h 32m39916
2Claude Fable 5.10.3524$77.13147.0M311k14h 51m4725
3Claude Fable 5.10.2715$77.56137.7M370k10h 32m4267
4Claude Fable 5.10.2491$104.18191.5M350k15h 56m5628
5Claude Fable 5.10.1812$69.87126.5M304k15h 13m4189
6GPT-5.60.1207$165.68160.5M163k9h 38m4167
7Grok 4.60.1206$114.8193.3M882k19h 41m42610
8Qwen3.8-Max0.1078$31.5667.3M1.0M19h 45m4562
9DeepSeek V4 Flash Vision Exp0.0671$3.79218.1M451k16h 2m5914
10Kimi K30.0612$51.38138.7M388k19h 16m4569
11Inkling0.0559$112.11117.1M218k1h 53m8111
12GLM-5.30.0490$52.08118.9M330k18h 59m3542
13DeepSeek V4 Flash Vision Exp0.0421$4.04235.7M470k19h 34m6055
14GLM-5.30.0406$79.08158.6M318k18h 36m4068
15Qwen3.8-Max0.0385$33.8372.9M764k18h 52m4222
16Kimi K30.0382$61.72163.7M377k19h 53m5343
17GLM-5.30.0363$101.89234.7M397k19h 55m4871
18Grok 4.60.0335$207.50166.7M1.1M18h 40m69223
19GLM-5.30.0278$99.55248.0M367k18h 54m56314
20Gemini 3.7 Flash0.0235$5.0624.8M145k2h 40m1971
21Qwen3.8-Max0.0216$32.9271.7M721k19h 16m4252
22Kimi K30.0211$44.44114.5M382k20h 0m3861
23Inkling0.0186$4.324.4M39k1h 1m1221
24GLM-5.30.0180$78.52153.1M353k18h 42m3969
25DeepSeek V4 Flash Vision Exp0.0178$3.85221.0M459k11h 59m5557
26Kimi K30.0143$59.95154.4M389k20h 1m5072
27Muse Spark 1.20.0135$59.82110.7M431k11h 54m3316
28Inkling0.0129$1.441.4M20k2h 0m841
29Muse Spark 1.20.0128$52.1470.4M490k14h 49m2894
30Gemini 3.7 Flash0.0127$18.6692.0M417k5h 34m3182
31Grok 4.60.0122$159.04128.5M975k18h 46m56920
32Gemini 3.7 Flash0.0121$11.5163.9M262k4h 49m3102
33Gemini 3.7 Flash0.0118$6.8126.5M177k3h 38m1832
34DeepSeek V4 Flash Vision Exp0.0114$4.73279.3M524k19h 31m6814
35Qwen3.8-Max0.0061$53.75151.0M965k20h 0m6952
36Muse Spark 1.20.0051$32.9057.9M354k9h 41m3723
37Grok 4.60.0043$218.28183.9M1.3M14h 37m75215
38Gemini 3.7 Flash0.0041$7.3932.0M185k3h 17m2131
39Inkling0.0031$5.445.4M67k1h 58m1202
40Inkling0.0025$2.422.4M30k2h 1m892
41DeepSeek V4 Flash Vision Exp0.0000$2.30127.9M331k19h 43m4543
42GPT-5.60.0000$58.1463.0M111k13h 6m2884
43GPT-5.60.0000$25.7538.4M104k16h 16m2184
44Grok 4.60.0000$57.5052.5M468k7h 59m2523
45Qwen3.8-Max0.0000$18.0251.8M431k20h 1m4332
46GPT-5.60.0000$51.9957.9M124k4h 17m2776
47GPT-5.60.0000$79.4380.8M113k10h 50m2854
48Kimi K30.0000$56.92159.4M376k20h 0m5211
49Muse Spark 1.20.0000$35.1065.9M265k9h 20m2293
50Muse Spark 1.20.0000$27.43124.8M316k10h 1m3854

† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.

Verilog Simulator in Swift

task page →50 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Claude Fable 5.10.9332$262.30478.6M947k9h 23m114817
2Claude Fable 5.10.9315$216.99391.4M878k7h 55m9666
3Claude Fable 5.10.9195$211.66382.5M840k9h 40m98917
4Claude Fable 5.10.9161$266.36483.7M1.0M10h 14m116734
5Claude Fable 5.10.9092$168.96308.4M606k6h 47m58610
6GPT-5.60.8853$427.16420.2M394k14h 37m9219
7GPT-5.60.8596$271.89260.9M315k4h 56m6505
8GPT-5.60.8493$335.11326.9M349k4h 23m7469
9Kimi K30.8373$248.18704.4M1.6M20h 0m163611
10GPT-5.60.8339$184.09180.1M273k4h 24m4806
11Gemini 3.7 Flash0.8305$103.651018.6M2.0M19h 58m24181
12Gemini 3.7 Flash0.8288$74.39745.3M2.1M17h 44m18281
13Grok 4.60.8253$223.65211.2M1.9M12h 53m88431
14GLM-5.30.8236$236.70879.4M1.3M20h 1m192411
15GPT-5.60.8236$181.11176.3M277k4h 52m4763
16Gemini 3.7 Flash0.8202$75.29725.4M1.4M19h 4m16512
17GLM-5.30.8151$195.25722.0M1.1M19h 28m170610
18Grok 4.60.8116$347.28325.2M2.2M19h 55m122256
19GLM-5.30.7757$229.93817.6M1.1M19h 47m180019
20GLM-5.30.7723$247.89875.5M1.2M19h 55m180417
21Kimi K30.7551$141.93387.6M1.1M20h 1m10091
22Qwen3.8-Max0.7551$162.17597.6M1.8M19h 37m14911
23GLM-5.30.7551$285.371061.6M1.4M19h 46m236723
24Grok 4.60.7517$518.10508.9M2.5M19h 53m184250
25Kimi K30.7175$256.44747.8M1.5M20h 1m16692
26Grok 4.60.6901$168.65160.9M1.6M9h 35m69125
27Kimi K30.6849$265.41774.0M1.6M19h 54m16822
28Kimi K30.6250$235.66685.5M1.4M19h 49m15211
29Gemini 3.7 Flash0.5976$41.43380.1M776k20h 1m9444
30DeepSeek V4 Flash Vision Exp0.5582$23.111453.0M1.8M19h 40m31183
31Grok 4.60.5188$107.0399.0M943k6h 25m40219
32Qwen3.8-Max0.4384$128.73471.1M1.6M19h 41m9652
33DeepSeek V4 Flash Vision Exp0.4212$16.08998.7M1.4M20h 0m21284
34DeepSeek V4 Flash Vision Exp0.3784$22.061380.9M1.8M19h 50m299912
35Gemini 3.7 Flash0.3545$53.22180.3M689k19h 59m5741
36Qwen3.8-Max0.3527$147.05546.0M1.5M19h 49m11092
37DeepSeek V4 Flash Vision Exp0.3253$27.961735.1M2.4M17h 46m379822
38DeepSeek V4 Flash Vision Exp0.3065$23.641484.7M1.9M18h 25m315730
39Qwen3.8-Max0.2894$161.74608.8M1.4M19h 56m14062
40Qwen3.8-Max0.2414$54.19188.0M1.1M20h 0m5971
41Muse Spark 1.20.0445$10.8558.8M321k1h 31m2292
42Muse Spark 1.20.0325$16.8184.3M312k1h 17m3292
43Muse Spark 1.20.0120$23.14130.5M679k1h 19m3953
44Muse Spark 1.20.0051$4.2721.5M175k51m1381
45Inkling0.0017$19.9320.8M47k1h 36m2831
46Inkling0.0000$3.273.4M20k1h 3m982
47Inkling0.0000$13.3613.9M43k1h 28m2271
48Muse Spark 1.20.0000$10.6954.0M468k1h 50m2081
49Inkling0.0000$2.172.2M13k1h 11m801
50Inkling0.0000$9.6410.0M34k1h 28m1692