Benchmarks/Aider polyglot

Aider polyglot

Aider Polyglot — measures how well AI models can edit code across multiple programming languages using the Aider coding assistant framework.

coding

Models Tested

88.0

Top Score

44.0

Average Score

Rankings

#	Model	Score
1	GPT-5 ChatOpenAI	88.0
2	GPT-5OpenAI	88.0
3	o3 ProOpenAI	84.9
4	Gemini 2.5 ProGoogle DeepMind	83.1
5	o3OpenAI	81.3
6	Grok 4xAI	79.6
7	DeepSeek V3.2DeepSeek	74.2
8	DeepSeek V3.2 ExpDeepSeek	74.2
9	Claude Opus 4Anthropic	72.0
10	o4 MiniOpenAI	72.0
11	DeepSeek-R1 (May 2025)DeepSeek	71.4
12	Claude 3.7 SonnetAnthropic	64.9
13	Claude 3.7 Sonnet (thinking)Anthropic	64.9
14	o1OpenAI	61.7
15	Claude Sonnet 4Anthropic	61.3
16	o3 MiniOpenAI	60.4
17	Qwen3 235B A22BAlibaba Qwen	59.6
18	Qwen3-235B-A22B-Instruct (Jul 2025)Alibaba	59.6
19	M Kimi K2 0905moonshotai	59.1
20	M Kimi K2 0711moonshotai	59.1
21	DeepSeek-R1DeepSeek	56.9
22	Grok 3xAI	53.3
23	Grok 3 BetaxAI	53.3
24	GPT-4.1OpenAI	52.4
25	Grok 3 MinixAI	49.3
26	Grok 3 Mini BetaxAI	49.3
27	DeepSeek V3DeepSeek	48.4
28	Gemini 2.5 FlashGoogle DeepMind	47.1
29	GPT-4.5OpenAI	44.9
30	gpt-oss-120b (free)OpenAI	41.8
31	gpt-oss-120bOpenAI	41.8
32	gpt-oss-20b (free)OpenAI	41.8
33	gpt-oss-20bOpenAI	41.8
34	Gemini 2.0 FlashGoogle DeepMind	38.2
35	Gemini 2.0 ProGoogle	35.6
36	o1-miniOpenAI	32.9
37	GPT-4.1 MiniOpenAI	32.4
38	Claude 3.5 HaikuAnthropic	28.0
39	GPT-4o (2024-08-06)OpenAI	23.1
40	GPT-4o (2024-05-13)OpenAI	23.1
41	GPT-4oOpenAI	23.1
42	GPT-4o (extended)OpenAI	23.1
43	Qwen2.5-MaxAlibaba	21.8
44	GPT-4o (2024-11-20)OpenAI	18.2
45	Gemini 2.0 Flash Thinking (Jan 2025)Google	18.2
46	Llama 4 MaverickMeta	15.6
47	GPT-4.1 NanoOpenAI	8.9
48	Gemma 3 4B (free)Google DeepMind	4.9
49	Gemma 3 4BGoogle DeepMind	4.9
50	Gemma 3 12B (free)Google DeepMind	4.9
51	Gemma 3 12BGoogle DeepMind	4.9
52	Gemma 3 27B (free)Google DeepMind	4.9
53	Gemma 3 27BGoogle DeepMind	4.9
54	GPT-4o-mini (2024-07-18)OpenAI	3.6
55	GPT-4o-miniOpenAI	3.6

Related Benchmarks

SWE-Bench Verified (Bash Only)

32 models

Terminal Bench

27 models