Skip to content
IFAI

Model rankings

1203 models · 5 boards · 2 sources

Artificial AnalysisIntelligence IndexArenaText arena

Text arena · Anonymous head-to-head votes · Published May 7, 2026

Most preferred
claude-opus-4-7-thinkingAnthropic1,503Score
Best open model
glm-5.1Z.ai1,471Score
Best from China
ernie-5.1Baidu1,474Score
Most voted
claude-3-opus-20240229Anthropic194,909Votes
Top three by Score
1claude-opus-4-7-thinkingAnthropic1,503
2claude-opus-4-6-thinkingAnthropic1,502
Top three open models
1glm-5.1Z.ai · #17 overall1,471
2mimo-v2.5-proXiaomi · #22 overall1,464
3deepseek-v4-proDeepSeek · #24 overall1,463
This week's changes
StepAudio 2.5 TTSSpeech · #3New
Bach-1.0 PreviewVideo · #7New
SIMBA 3.0Speech · #7New
qwen3.6-max-previewCoding · #9New
RankChangeModelScore95% intervalVotesPrice $/M
1—ernie-5.1ClosedBaidu1,474±85,733—
2—glm-5.1OpenZ.ai1,471±611,349—
3—mimo-v2.5-proOpenXiaomi1,464±76,238—
4▲ 1qwen3.5-max-previewClosedAlibaba1,464±514,558—
5▼ 1deepseek-v4-proOpenDeepSeek · 1 more variant1,463±94,160—
6—kimi-k2.6OpenMoonshot AI1,462±77,108—
7—qwen3.6-max-previewClosedAlibaba1,457±93,965—
8—glm-5OpenZ.ai1,457±520,292—
9—ernie-5.0-0110ClosedBaidu1,450±428,724—
10—kimi-k2.5-thinkingOpenMoonshot AI1,449±427,282—
11—ernie-5.0-preview-1203ClosedBaidu · 1 more variant1,449±79,764—
12▲ 1qwen3.6-plusClosedAlibaba1,448±68,683—
13▼ 1mimo-v2-proClosedXiaomi1,447±515,257—
14—qwen3.5-397b-a17bOpenAlibaba1,446±522,471—
15—glm-4.7OpenZ.ai1,443±612,142—
16—deepseek-v4-flash-thinkingOpenDeepSeek · 1 more variant1,440±93,600—
17—qwen3-max-previewClosedAlibaba1,435±427,743—
18—longcat-flash-chat-2602-expClosedMeituan1,434±613,311—
19—kimi-k2.5-instantOpenMoonshot AI1,432±78,207—
20—kimi-k2-thinking-turboOpenMoonshot AI1,430±352,935—
21—glm-4.6OpenZ.ai1,426±435,694—
22—deepseek-v3.2-exp-thinkingOpenDeepSeek · 1 more variant1,425±79,076—
23▲ 1deepseek-v3.2OpenDeepSeek · 1 more variant1,424±444,820—
24▲ 1qwen3-max-2025-09-23ClosedAlibaba1,424±69,179—
25▼ 2mimo-v2.5OpenXiaomi1,423±76,300—
26—qwen3-235b-a22b-instruct-2507OpenAlibaba1,423±388,518—
27—deepseek-r1-0528OpenDeepSeek · 1 more variant1,422±618,469—
28▲ 1qwen3.5-122b-a10bOpenAlibaba1,418±519,379—
29▼ 1hunyuan-hy3-previewOpenTencent1,418±84,582—
30—kimi-k2-0905-previewOpenMoonshot AI · 1 more variant1,418±611,798—
31—deepseek-v3.1OpenDeepSeek · 1 more variant1,418±614,985—
32—deepseek-v3.1-terminus-thinkingOpenDeepSeek · 1 more variant1,417±103,474—
33—qwen3-vl-235b-a22b-instructOpenAlibaba1,415±611,529—
34—glm-4.5OpenZ.ai1,411±524,336—
35▲ 1minimax-m2.7OpenMiniMax1,407±613,525—
36▼ 1qwen3.5-27bOpenAlibaba1,406±518,942—
37—qwen3-235b-a22b-no-thinkingOpenAlibaba1,403±438,241—
38—qwen3-next-80b-a3b-instructOpenAlibaba1,402±522,883—
39—longcat-flash-chatOpenMeituan1,401±611,409—
40—qwen3-235b-a22b-thinking-2507OpenAlibaba1,399±79,004—
41—qwen3.5-flashClosedAlibaba1,398±519,649—
42▲ 3qwen3.5-35b-a3bOpenAlibaba1,397±519,774—
43—hunyuan-vision-1.5-thinkingClosedTencent1,396±122,221—
44—qwen3-vl-235b-a22b-thinkingOpenAlibaba1,396±77,944—
45▲ 1deepseek-v3-0324OpenDeepSeek · 1 more variant1,395±445,533—
46▼ 4minimax-m2.5OpenMiniMax1,395±424,885—
47▲ 1step-3.5-flashOpenStepFun1,393±425,112—
48▼ 1mimo-v2-flash (non-thinking)OpenXiaomi · 1 more variant1,393±437,247—
49▲ 1hunyuan-t1-20250711ClosedTencent1,387±94,710—
50▼ 1qwen3-coder-480b-a35b-instructOpenAlibaba1,387±525,757—

Intelligence over time

Each maker's best model at the time on the Artificial Analysis Intelligence Index, by the month it was released.

Gained most · Anthropic +57.2Highest now · Anthropic 64.6Gap between highest and lowest 18.1 ptsTime range Nov 2025 – Nov 2026
64.6Anthropic60.0Google56.9OpenAI50.1Meta46.5Kimi

Arena data comes from the leaderboard dataset it publishes on Hugging Face (CC BY 4.0). Only each board's overall leaderboard is used; scores and ranks are unchanged. dataset · CC BY 4.0

Artificial Analysis data comes from its API (artificialanalysis.ai). Scores are unchanged; ranks follow the scores. artificialanalysis.ai

Open or closed follows the license Arena gives; prices and Artificial Analysis intelligence show only where model names match, otherwise “—”.