RankingUpdates

LLM updates: week to 22 Nov 2026

Updated · Evidence published · This page updates each day from the published ranking.

4 models were released in the last 7 days: Gemini 4 Neon, Astra for Law 6.2, GPT-6.2 Astra and GPT-6.2 Sol. GPT-6 Astra · XHigh is #1 on UnifyBench at 95.3.

Which LLM is #1 this week?

The latest checks show different measures of performance: GPT-6 Astra · XHigh leads UnifyBench; Claude Fable 5.2 (Max, Default Fallback) leads Artificial Analysis; Gemini 4 Carbon (High) leads Arena. Shortlist with UnifyBench, check Artificial Analysis for independent scores and cost per task, check Arena for preferences, then test your own tasks.

The Intelligence Index, Arena Elo, and UnifyBench score use different scales and cannot be compared directly. Why leaderboards disagree.

Which new LLMs were released this week?

Releases in the last 30 days

Model releases of the last 30 days with their rank on three boards
ModelLabReleasedUnifyBench rankUnifyBench scoreAA Intelligence IndexArena text rank
Gemini 4 NeonGoogle#690.061 (#3)#3 · Elo 1513
Astra for Law 6.2OpenAIEvidence pending—No index yetNot rated yet
GPT-6.2 AstraOpenAI#390.762 (#2)#55
GPT-6.2 SolOpenAI#1379.657 (#6)#64
Gemini 4 HydrogenGoogle#1182.554 (#8)#18
Claude Fable 5.2Anthropic#788.865 (#1)#7 · Elo 1503
Muse EmberMeta#3553.236 (#16)Not rated yet
Agnes 2.5 Pro ThetaAgnes AI#5835.124 (#33)Not rated yet
Tencent Hy4Tencent#2167.935 (#18)Not rated yet
Gemini 4 CarbonGoogle#291.660 (#4)#1 · Elo 1527
GPT-6.1 TerraOpenAI#1777.451 (#9)#44

UnifyBench rank and score: default Overall, one entry per model on its highest-scoring effort setting. "Evidence pending" means that no published benchmark result supports a rank yet. The Artificial Analysis index is the highest listed setting of the release, checked 22 Nov 2026. The Arena rank is the best listed setting of the model, checked 22 Nov 2026.

Top 10 rank changes

Rank changes appear after two published days of rank history.

Full leaderboard · Methodology · Data