Best AI models for coding, November 9, 2026

The top 30 models on Arena's web development leaderboard: people see two generated web apps and vote for the better one.

Key facts

First: Claude Opus 5.5 (Max) (1,818), then Claude Sonnet 5.5 (xHigh) (1,774) and Gemini 4 Carbon (High) (1,768).
Anthropic holds the most places in the top 10: 5.
Data as of November 9, 2026.

Download CSV

30 rows

Best AI models for coding, November 9, 2026. Select a column heading to sort.

1	Claude Opus 5.5 (Max)	Anthropic	Proprietary	1,818	4,009
2	Claude Sonnet 5.5 (xHigh)	Anthropic	Proprietary	1,774	5,493
3	Gemini 4 Carbon (High)	Google	Proprietary	1,768	2,128
4	GPT-6 Astra (Max)	OpenAI	Proprietary	1,767	8,117
5	Claude Fable 5.1 (Max)	Anthropic	Proprietary	1,745	9,453
6	GPT-6.1 Sol (Max)	OpenAI	Proprietary	1,743	5,250
7	Muse Wonder Full (Max)	Meta	Proprietary	1,724	2,080
8	Claude Sonnet 5.5 (High)	Anthropic	Proprietary	1,713	7,541
9	Claude Opus 5 (Max)	Anthropic	Proprietary	1,691	19,593
10	Kimi K3.1 Pro (Max)	Moonshot AI	Kimi K3.1 Pro License	1,688	4,591
11	Gemini 4 Argon (High)	Google	Proprietary	1,683	4,626
12	GPT-6.1 Terra (Max)	OpenAI	Proprietary	1,679	3,559
13	Qwen3.8-Max (0902)	Alibaba	Proprietary	1,677	21,354
14	GPT-6 Sol (Max)	OpenAI	Proprietary	1,674	6,882
15	Qwen3.8-Max	Alibaba	Proprietary	1,670	3,373
16	DeepSeek-V4.1-Flash-20261017 (Max)	DeepSeek	MIT	1,664	13,241
17	Kimi K3 (Max)	Moonshot AI	Kimi K3 license	1,657	22,114
18	Claude Opus 5 (High)	Anthropic	Proprietary	1,657	26,982
19	Muse Spark 1.3 (Max)	Meta	Proprietary	1,654	11,758
20	Muse Wonder (Max)	Meta	Proprietary	1,650	4,875
21	Inkling 2 Small	Thinky	Proprietary	1,649	964
22	Grok 4.7 (xHigh)	xAI	Proprietary	1,639	7,816
23	MiMo-V2.6-Flash	Xiaomi	MIT	1,637	8,576
24	Hy4 preview	Tencent	Apache 2.0	1,631	4,606
25	Claude Fable 5 (High)	Anthropic	Proprietary	1,625	17,017
26	Qwen3.8-Flash-Next	Alibaba	qwen-community-1.0	1,624	10,939
27	MiMo-V2.6-Pro	Xiaomi	MIT	1,624	8,402
28	GLM-5.3 (Max)	Z.ai	MIT	1,622	12,263
29	DeepSeek-V4.1-Flash-20260910 (Max)	DeepSeek	MIT	1,621	3,408
30	Muse Spark 1.3 (xHigh)	Meta	Proprietary	1,621	11,277

Source: Arena leaderboard dataset · Licence: CC BY 4.0 · Data as of November 9, 2026 · Collected 2026-11-10

About this ranking

Each vote compares two anonymous models given the same request to build a web page or small app. Voters judge the result they see, so this table rewards working, good-looking front ends more than back-end or algorithm skill.

Scores are ratings built from those votes. Models with few votes can move a lot, so read the votes column too, and test the top few on your own codebase, which no public ranking can stand in for.

FAQ

What is the best AI for coding right now?

In Arena's web development votes as of November 9, 2026, Claude Opus 5.5 (Max) ranks first with 1,818, then Claude Sonnet 5.5 (xHigh) and Gemini 4 Carbon (High).

Does this cover all kinds of programming?

No. It covers building web pages and apps from a prompt. A model that wins here may not be the best at debugging a large project or writing systems code.

Why are the scores higher than in the chat table?

Each table is rated from its own votes, so a score only compares models within one table.

More rankings

All rankings