Pass IndexThe State of AISign in

Aider polyglot — Epoch AI

Aider polyglot — Epoch AI is run by Epoch AI. It has ranked 72 models, of which the catalogue holds 46, scoring from 3.6 to 88.

measured by Epoch AI · the board itself

PlaceModelMetricScore
1stof 72GPT-5Percent correct (high)88
3rdof 72o3-proPercent correct (high)84.9
4thof 72Gemini 2.5 Pro Preview 06-05Percent correct83.1
6thof 72o3Percent correct (high)81.3
11thof 72Gemini 2.5 Pro Preview 05-06Percent correct76.9
14thof 72deepseek-reasonerPercent correct74.2
16thof 72Gemini 2.5 ProPercent correct72.9
17thof 72Claude Opus 4Percent correct72
18thof 72o4-miniPercent correct (high)72
19thof 72DeepSeek-R1 (0528)Percent correct71.4
21stof 72DeepSeek V3.2 ExpPercent correct70.2
22ndof 72deepseek-chatPercent correct70.2
23rdof 72Claude 3.7 SonnetPercent correct64.9
24thof 72o1Percent correct (high)61.7
25thof 72Claude Sonnet 4Percent correct61.3
27thof 72o3-miniPercent correct (high)60.4
28thof 72Qwen3 235B A22BPercent correct59.6
29thof 72Qwen3-235B-A22B-Instruct-2507Percent correct59.6
30thof 72Kimi K2 InstructPercent correct59.1
31stof 72Kimi K2 0905Percent correct59.1
32ndof 72DeepSeek-R1Percent correct56.9
34thof 72Gemini 2.5 FlashPercent correct55.1
35thof 72DeepSeek-V3 0324Percent correct55.1
37thof 72Grok 3Percent correct53.3
38thof 72GPT-4.1 fine-tunedPercent correct52.4
39thof 72Claude 3.5 SonnetPercent correct51.6
40thof 72Grok 3 MiniPercent correct (high)49.3
41stof 72DeepSeek-V3Percent correct48.4
46thof 72GPT OSS 120BPercent correct (high)41.8
48thof 72Qwen3 32BPercent correct40
52ndof 72o1-miniPercent correct32.9
53rdof 72GPT-4.1 miniPercent correct32.4
54thof 72Claude Haiku 3.5Percent correct28
56thof 72GPT-4o (2024-08-06)Percent correct23.1
57thof 72Gemini 2.0 FlashPercent correct22.2
59thof 72QwQ-32BPercent correct20.9
61stof 72GPT-4o (2024-11-20)Percent correct18.2
62ndof 72DeepSeek-V2.5Percent correct17.8
63rdof 72Qwen2.5 Coder 32B InstructPercent correct16.4
64thof 72Llama-4-Maverick-17B-128E-InstructPercent correct15.6
66thof 72c4ai-command-a-03-2025Percent correct12
67thof 72Codestral 25.01Percent correct11.1
68thof 72openhands-lm-32b-v0.1Percent correct10.2
69thof 72GPT-4.1 nanoPercent correct8.9
71stof 72Gemma 3 27BPercent correct4.9
72ndof 72GPT-4o-mini (2024-07-18)Percent correct3.6

Read from the board on 2026-09-16