Code Arena

Compare the performance of AI models on agentic coding tasks involving multi-step reasoning and tool use

Last Updated

Jan 29, 2026

Total Votes

120,051

Total Models

/

	Rank Spread
1	1◄─►1	claude-opus-4-5-20251101-thinking-32k	1497	+10/-10	7,953	Anthropic	Proprietary
2	2◄─►4	gpt-5.2-high	1470	+16/-16	1,689	OpenAI	Proprietary
3	2◄─►4	claude-opus-4-5-20251101	1468	+9/-9	8,193	Anthropic	Proprietary
4	2◄─►6	gemini-3-pro	1454	+8/-8	14,107	Google	Proprietary
5	4◄─►6	gemini-3-flash	1443	+9/-9	9,548	Google	Proprietary
6	4◄─►6	glm-4.7	1440	+10/-10	5,110	Z.ai	MIT
7	7◄─►10	minimax-m2.1-preview	1408	+9/-9	7,161	MiniMax	MIT
8	7◄─►14	gemini-3-flash (thinking-minimal)	1399	+10/-10	5,612	Google	Proprietary
9	7◄─►15	gpt-5.2	1395	+16/-16	1,632	OpenAI	Proprietary
10	7◄─►15	gpt-5-medium	1392	+12/-12	3,926	OpenAI	Proprietary
11	8◄─►15	claude-opus-4-1-20250805	1387	+9/-9	8,975	Anthropic	Proprietary
12	8◄─►15	gpt-5.1-medium	1387	+9/-9	6,429	OpenAI	Proprietary
13	8◄─►15	claude-sonnet-4-5-20250929	1386	+8/-8	12,959	Anthropic	Proprietary
14	8◄─►15	claude-sonnet-4-5-20250929-thinking-32k	1386	+8/-8	11,367	Anthropic	Proprietary
15	9◄─►16	deepseek-v3.2-thinking	1372	+11/-11	3,782	DeepSeek	MIT
16	15◄─►18	glm-4.6	1355	+9/-9	8,738	Z.ai	MIT
17	16◄─►19	gpt-5.1	1351	+8/-8	10,303	OpenAI	Proprietary
18	16◄─►21	mimo-v2-flash (non-thinking)	1347	+10/-10	4,223	Xiaomi	MIT
19	17◄─►22	gpt-5.2-codex	1331	+12/-12	2,919	OpenAI	Proprietary
20	18◄─►21	kimi-k2-thinking-turbo	1329	+8/-8	9,891	Moonshot	Modified MIT
21	18◄─►22	gpt-5.1-codex	1328	+9/-9	6,501	OpenAI	Proprietary
22	20◄─►23	minimax-m2	1311	+9/-9	8,838	MiniMax	Apache 2.0
23	22◄─►26	deepseek-v3.2	1296	+10/-10	4,844	DeepSeek	MIT
24	23◄─►26	claude-haiku-4-5-20251001	1292	+8/-8	11,111	Anthropic	Proprietary
25	23◄─►26	deepseek-v3.2-exp	1285	+10/-10	5,131	DeepSeek	MIT
26	23◄─►26	qwen3-coder-480b-a35b-instruct	1282	+8/-8	10,859	Alibaba	Apache 2.0
27	27◄─►29	KAT-Coder-Pro-V1	1258	+15/-15	1,956	KwaiKAT	Proprietary
28	27◄─►30	gpt-5.1-codex-mini	1242	+17/-17	1,537	OpenAI	Proprietary
29	27◄─►30	grok-4-1-fast-reasoning	1236	+10/-10	5,682	xAI	Proprietary
30	28◄─►33	mistral-large-3	1221	+20/-20	1,039	Mistral	Apache 2.0
31	30◄─►33	gemini-2.5-pro	1204	+13/-13	3,454	Google	Proprietary
32	30◄─►33	grok-4.1-thinking	1203	+19/-19	1,266	xAI	Proprietary
33	30◄─►33	devstral-2	1201	+16/-16	1,659	Mistral	Modified MIT
34	34◄─►35	grok-4-fast-reasoning	1151	+22/-22	970	xAI	Proprietary
35	34◄─►36	grok-code-fast-1	1139	+21/-21	1,017	xAI	Proprietary
36	35◄─►36	devstral-medium-2507	1097	+22/-22	1,020	Mistral	Proprietary

Code Arena

Remove Style Control Leaderboard Plots

Fraction of Model A Wins for All Non-tied A vs. B Battles

Battle Count for Each Combination of Models (without Ties)

Confidence Intervals on Model Strength (via Bootstrapping)

Average Win Rate Against All Other Models (Uniform Sampling and No Ties)