Code Arena | WebDev

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

Feb 19, 2026

163,881 votes

45 models

	Rank Spread
1	12	claude-opus-4-6 Anthropic · Proprietary	1561+14/-14	2,524
2	13	claude-opus-4-6-thinking Anthropic · Proprietary	1551+16/-16	1,919
3	24	claude-sonnet-4-6 Anthropic · Proprietary	1524+20/-20	1,021
4	34	claude-opus-4-5-20251101-thinking-32k Anthropic · Proprietary	1501+8/-8	10,620
5	59	gpt-5.2-high OpenAI · Proprietary	1471+16/-16	1,695
6	58	claude-opus-4-5-20251101 Anthropic · Proprietary	1469+8/-8	10,730
7	513	gemini-3.1-pro-preview Google · Proprietary	1461+15/-15	1,831
8	513	glm-5 Z.ai · MIT	1456+14/-14	2,243
9	714	gemini-3-pro Google · Proprietary	1444+7/-7	16,673
10	614	minimax-m2.5 MiniMax · Modified MIT	1443+12/-12	3,260
11	714	glm-4.7 Z.ai · MIT	1440+10/-10	5,128
12	714	gemini-3-flash Google · Proprietary	1440+8/-8	12,350
13	714	kimi-k2.5-thinking Moonshot · Modified MIT	1439+11/-11	3,564
14	914	kimi-k2.5-instant Moonshot · Modified MIT	1424+13/-13	2,482
15	1521	minimax-m2.1-preview MiniMax · MIT	1402+8/-8	9,526
16	1521	gemini-3-flash (thinking-minimal) Google · Proprietary	1402+8/-8	8,383
17	1522	gpt-5.2 OpenAI · Proprietary	1395+16/-15	1,634
18	1522	gpt-5-medium OpenAI · Proprietary	1393+12/-12	3,928
19	1522	claude-sonnet-4-5-20250929-thinking-32k Anthropic · Proprietary	1390+7/-7	13,816
20	1522	claude-opus-4-1-20250805 Anthropic · Proprietary	1389+8/-8	8,985
21	1523	gpt-5.1-medium OpenAI · Proprietary	1388+9/-9	6,438
22	1723	claude-sonnet-4-5-20250929 Anthropic · Proprietary	1386+7/-7	15,469
23	2124	deepseek-v3.2-thinking DeepSeek · MIT	1371+9/-9	5,714
24	2326	glm-4.6 Z.ai · MIT	1356+8/-8	8,746
25	2429	gpt-5.1 OpenAI · Proprietary	1343+7/-7	12,764
26	2429	mimo-v2-flash (non-thinking) Xiaomi · MIT	1342+8/-8	6,664
27	2530	gpt-5.2-codex OpenAI · Proprietary	1336+9/-9	5,380
28	2530	kimi-k2-thinking-turbo Moonshot · Modified MIT	1331+7/-7	12,256
29	2532	gpt-5.1-codex OpenAI · Proprietary	1329+9/-9	6,506
30	2733	deepseek-v3.2 DeepSeek · MIT	1319+9/-9	7,006
31	2933	minimax-m2 MiniMax · Apache 2.0	1312+9/-9	8,834
32	3033	claude-haiku-4-5-20251001 Anthropic · Proprietary	1307+7/-7	13,542
33	2934	mimo-v2-flash (thinking) Xiaomi · MIT	1306+13/-13	2,143
34	3335	deepseek-v3.2-exp DeepSeek · MIT	1286+10/-10	5,131
35	3435	qwen3-coder-480b-a35b-instruct Alibaba · Apache 2.0	1282+7/-7	13,266
36	3638	KAT-Coder-Pro-V1 KwaiKAT · Proprietary	1259+15/-15	1,954
37	3639	gpt-5.1-codex-mini OpenAI · Proprietary	1243+17/-17	1,537
38	3639	grok-4-1-fast-reasoning xAI · Proprietary	1236+9/-9	7,129
39	3742	mistral-large-3 Mistral · Apache 2.0	1223+20/-20	1,039
40	3942	gemini-2.5-pro Google · Proprietary	1205+13/-13	3,454
41	3942	grok-4.1-thinking xAI · Proprietary	1204+19/-19	1,267
42	3942	devstral-2 Mistral · Modified MIT	1199+16/-16	1,683
43	4344	grok-4-fast-reasoning xAI · Proprietary	1153+22/-22	968
44	4345	grok-code-fast-1 xAI · Proprietary	1141+21/-21	1,017
45	4445	devstral-medium-2507 Mistral · Proprietary	1099+22/-22	1,020

Code Arena | WebDev

Remove Style Control Leaderboard Plots

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)

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