LMArena Code Leaderboard (September 2026)
In September 2026, claude-fable-5.1-max ranks first with 1,765, followed by qwen3.8-max-0902 (1,688) and claude-opus-5-max (1,687). Top 20 of 118 candidates.
Data as of September 2026
#ModelProviderScoreVotes
1Cclaude-fable-5.1-maxAnthropic1,7651.1K
2Qqwen3.8-max-0902Alibaba1,6881.5K
3Cclaude-opus-5-maxAnthropic1,68710.6K
4KKimi K3Moonshot1,6791.8K
5Kkimi-k3-maxMoonshot1,6744.5K
6QQwen 3.8 MaxAlibaba1,6693.2K
7Cclaude-opus-5-highAnthropic1,66110.5K
8Ggrok-4.6-highSpaceXAI1,6291.5K
9CClaude Fable 5Anthropic1,6289.2K
10Hhy4-previewTencent1,6261.5K
11Qqwen3.8-flash-nextAlibaba1,6222.0K
12Ggpt-5.6-sol-xhigh (codex-harness)OpenAI1,61610.8K
13Gglm-5.3-maxZ.ai1,6092.7K
14Gglm-5.3-flashZ.ai1,6041.8K
15QQwen 3.8 27BAlibaba1,5973.3K
16Gglm-5.2 (max)Z.ai1,5874.7K
17Ggemini-3.7-flash-highGoogle1,5873.0K
18Gglm-5.2-maxZ.ai1,5859.7K
19Ddeepseek-v4-pro-high-20260813DeepSeek1,5823.4K
20Ddeepseek-v4-flash-highDeepSeek1,5814.0K
How this chart is built
From a public blind human-preference arena leaderboard; score and vote count are that leaderboard’s current snapshot. Also replaced wholesale each release, with no history.
Positions are renumbered from high to low by this table's own metric, independent of any ranking the source publishes. Figures are the current real data with no manual adjustment; values we cannot obtain show as “—” and are never filled in with estimates.