Skip to content
Weights Not Walls
Menu

MirrorCode

Re-run by Epoch AI

No description yet: we only describe an eval set when we can do it correctly.

Kind
Run independently by Epoch AI, with public logs. Our own re-runs are not built yet.
Score shown
Epoch's column "Best score (across scorers)", as a percentage, higher is better; random guessing scores about 0.0%. Error bars as the source gives them.
Results
47 in all: 5 for open-weight models, 42 for closed ones. 4 linked to a model page here.

Best open-weight result: deepseek-ai/DeepSeek-V4.1-Flash, 64.4%, number 12 of 47 results.

MirrorCode results
#ModelWeightsSettingScoreDateSource
1claude-sonnet-5-5_maxAnthropicclosedeffort max97.8%±1.55 Oct 2026Epoch AI log
2claude-opus-5-5_maxAnthropicclosedeffort max95.6%±2.630 Oct 2026Epoch AI
3claude-fable-5-1_maxAnthropicclosedeffort max93.3%±3.99 Sep 2026Epoch AI log
4claude-opus-5_maxAnthropicclosedeffort max87.8%±7.220 Oct 2026Epoch AI log
5claude-opus-5-5_highAnthropicclosedeffort high78.9%±7.922 Sep 2026Epoch AI
6claude-haiku-5-5_maxAnthropicclosedeffort max78.9%±8.021 Oct 2026Epoch AI log
7gpt-6.1-terra_maxOpenAIclosedeffort max76.7%±9.330 Oct 2026Epoch AI
8claude-fable-5-1_highAnthropicclosedeffort high73.3%±9.210 Sep 2026Epoch AI
9claude-fable-5_maxAnthropicclosedeffort max70.0%±10.430 Oct 2026Epoch AI
10gpt-6.2-astra_maxOpenAIclosedeffort max67.8%±11.727 Oct 2026Epoch AI
11claude-opus-4-8_maxAnthropicclosedeffort max65.6%±10.821 Oct 2026Epoch AI log
12deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_max, DeepSeekopeneffort max64.4%±11.421 Oct 2026Epoch AI
13claude-fable-5_highAnthropicclosedeffort high63.9%±10.410 Aug 2026Epoch AI
14claude-sonnet-5_maxAnthropicclosedeffort max63.3%±10.721 Oct 2026Epoch AI
15gpt-6.1-sol_maxOpenAIclosedeffort max61.1%±11.35 Oct 2026Epoch AI log
16gpt-6-astra_maxOpenAIclosedeffort max60.0%±11.829 Oct 2026Epoch AI log
17gemini-4-argon_highGoogle DeepMindclosedeffort high58.9%±11.714 Oct 2026Epoch AI log
18grok-4.7_xhighxAIclosedeffort xhigh54.4%±11.429 Oct 2026Epoch AI
19moonshotai/Kimi-K3Epoch: kimi-k3_max, Moonshotopeneffort max54.4%±11.43 Sep 2026Epoch AI log
20muse-spark-1.3_maxMeta AIclosedeffort max54.4%±10.820 Oct 2026Epoch AI
21qwen3.8-max_xhighAlibabaclosedeffort xhigh52.2%±10.921 Oct 2026Epoch AI log
22gpt-6-sol_maxOpenAIclosedeffort max48.9%±11.229 Oct 2026Epoch AI
23qwen3.8-max-0902_xhighAlibabaclosedeffort xhigh47.8%±10.321 Oct 2026Epoch AI log
24gpt-6-astra_highOpenAIclosedeffort high46.7%±11.930 Aug 2026Epoch AI
25muse-wonder-1.0_maxMeta AIclosedeffort max45.6%±10.614 Oct 2026Epoch AI log
26claude-opus-4-7_maxAnthropicclosedeffort max43.3%±9.630 Oct 2026Epoch AI
27gemini-3.7-flash_highGoogle DeepMindclosedeffort high41.1%±12.221 Oct 2026Epoch AI log
28gpt-5.6-terra_maxOpenAIclosedeffort max41.1%±12.420 Oct 2026Epoch AI log
29grok-4.6_xhighxAIclosedeffort xhigh41.1%±11.021 Oct 2026Epoch AI log
30mistral-large-4-0_highMistral AIno page matchopeneffort high40.0%±10.62 Nov 2026Epoch AI
31gemini-3.8-flash_highGoogle DeepMindclosedeffort high38.9%±11.92 Sep 2026Epoch AI log
32zai-org/GLM-5.2Epoch: glm-5.2_max, Z.ai (Zhipu AI)openeffort max36.7%±11.521 Oct 2026Epoch AI
33claude-opus-4-6_maxAnthropicclosedeffort max33.3%±10.021 Oct 2026Epoch AI log
34deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_max, DeepSeekopeneffort max33.3%±10.020 Oct 2026Epoch AI log
35gpt-5.2-2025-12-11_xhighOpenAIclosedeffort xhigh32.2%±10.021 Oct 2026Epoch AI
36grok-4.5_highxAIclosedeffort high32.2%±10.721 Oct 2026Epoch AI log
37claude-opus-4-7_highAnthropicclosedeffort high31.1%±9.012 Aug 2026Epoch AI
38gpt-6-luna_maxOpenAIclosedeffort max26.7%±9.021 Oct 2026Epoch AI log
39gemini-3.6-flash_highGoogle DeepMindclosedeffort high23.3%±9.321 Oct 2026Epoch AI
40gpt-5.4-2026-03-05_xhighOpenAIclosedeffort xhigh23.3%±8.129 Oct 2026Epoch AI
41gpt-5.5_xhighOpenAIclosedeffort xhigh23.3%±10.129 Oct 2026Epoch AI
42gpt-5.6-luna_maxOpenAIclosedeffort max21.1%±9.821 Oct 2026Epoch AI
43gpt-5.6-sol_highOpenAIclosedeffort high20.0%±9.210 Aug 2026Epoch AI
44gpt-5.4-2026-03-05_highOpenAIclosedeffort high15.6%±7.710 Aug 2026Epoch AI
45gpt-5.5_highOpenAIclosedeffort high10.0%±6.010 Aug 2026Epoch AI
46gemini-3.1-pro-preview_highGoogle DeepMindclosedeffort high8.9%±4.610 Aug 2026Epoch AI
47gpt-5-2025-08-07_highOpenAIclosedeffort high4.4%±3.421 Oct 2026Epoch AI log

Open-weight rows are shaded. Dates are when the result was run or evaluated, where the data says. Data from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Epoch AI. Method