Code Arena | WebDev

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

Feb 23, 2026

169,950 votes

46 models

	Rank Spread
1	13	claude-opus-4-6 Anthropic · Proprietary	1560+14/-14	2,766
2	13	claude-opus-4-6-thinking Anthropic · Proprietary	1553+15/-15	2,115
3	13	claude-sonnet-4-6 Anthropic · Proprietary	1533+16/-16	1,675
4	44	claude-opus-4-5-20251101-thinking-32k Anthropic · Proprietary	1499+8/-8	11,032
5	58	gpt-5.2-high OpenAI · Proprietary	1471+16/-16	1,696
6	58	claude-opus-4-5-20251101 Anthropic · Proprietary	1470+8/-8	11,113
7	513	gemini-3.1-pro-preview Google · Proprietary	1461+15/-15	1,826
8	513	glm-5 Z.ai · MIT	1452+13/-13	2,520
9	713	gemini-3-pro Google · Proprietary	1444+7/-7	16,948
10	713	gemini-3-flash Google · Proprietary	1440+8/-8	12,778
11	714	glm-4.7 Z.ai · MIT	1439+10/-10	5,127
12	714	minimax-m2.5 MiniMax · Modified MIT	1438+11/-11	3,557
13	714	kimi-k2.5-thinking Moonshot · Modified MIT	1436+10/-10	3,900
14	1118	kimi-k2.5-instant Moonshot · Modified MIT	1419+12/-12	2,839
15	1422	minimax-m2.1-preview MiniMax · MIT	1402+8/-8	9,796
16	1423	gemini-3-flash (thinking-minimal) Google · Proprietary	1400+8/-8	8,742
17	1423	qwen3.5-397b-a17b Alibaba · Apache 2.0	1396+13/-13	2,388
18	1423	gpt-5.2 OpenAI · Proprietary	1395+16/-15	1,634
19	1523	gpt-5-medium OpenAI · Proprietary	1393+12/-12	3,929
20	1523	claude-sonnet-4-5-20250929-thinking-32k Anthropic · Proprietary	1388+7/-7	14,117
21	1523	claude-opus-4-1-20250805 Anthropic · Proprietary	1388+8/-8	8,985
22	1524	gpt-5.1-medium OpenAI · Proprietary	1387+9/-9	6,438
23	1623	claude-sonnet-4-5-20250929 Anthropic · Proprietary	1386+7/-7	15,814
24	2325	deepseek-v3.2-thinking DeepSeek · MIT	1370+9/-9	5,960
25	2427	glm-4.6 Z.ai · MIT	1356+8/-8	8,747
26	2530	gpt-5.1 OpenAI · Proprietary	1343+7/-7	13,086
27	2530	mimo-v2-flash (non-thinking) Xiaomi · MIT	1341+8/-8	6,932
28	2630	gpt-5.2-codex OpenAI · Proprietary	1336+9/-9	5,708
29	2631	kimi-k2-thinking-turbo Moonshot · Modified MIT	1331+7/-7	12,589
30	2633	gpt-5.1-codex OpenAI · Proprietary	1328+9/-9	6,506
31	2934	deepseek-v3.2 DeepSeek · MIT	1318+8/-8	7,291
32	3034	minimax-m2 MiniMax · Apache 2.0	1312+9/-9	8,834
33	3035	mimo-v2-flash (thinking) Xiaomi · MIT	1306+13/-13	2,146
34	3134	claude-haiku-4-5-20251001 Anthropic · Proprietary	1306+7/-7	13,857
35	3436	deepseek-v3.2-exp DeepSeek · MIT	1286+10/-10	5,131
36	3537	qwen3-coder-480b-a35b-instruct Alibaba · Apache 2.0	1280+7/-7	13,588
37	3639	KAT-Coder-Pro-V1 KwaiKAT · Proprietary	1258+15/-15	1,955
38	3740	gpt-5.1-codex-mini OpenAI · Proprietary	1242+17/-17	1,537
39	3740	grok-4-1-fast-reasoning xAI · Proprietary	1235+9/-9	7,129
40	3843	mistral-large-3 Mistral · Apache 2.0	1222+20/-20	1,039
41	4043	gemini-2.5-pro Google · Proprietary	1205+13/-13	3,455
42	4043	grok-4.1-thinking xAI · Proprietary	1204+19/-19	1,267
43	4043	devstral-2 Mistral · Modified MIT	1197+16/-16	1,686
44	4445	grok-4-fast-reasoning xAI · Proprietary	1152+22/-22	968
45	4446	grok-code-fast-1 xAI · Proprietary	1140+21/-21	1,017
46	4546	devstral-medium-2507 Mistral · Proprietary	1099+22/-22	1,020

Code Arena | WebDev

Remove Style Control Leaderboard Plots

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

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

Confidence Intervals on Model Strength (via Bootstrapping)

Battle Count for Each Combination of Models (without Ties)