Atlas

Benchmarks

← All benchmarks

LMArena Text Arena Occupational Mathematical Elo

Math · 2025-11-05

Category-specific Elo score from the LMArena text-arena voting leaderboard. Higher Elo indicates stronger crowd-voted preference within this category.

Top models (higher is better)

ModelScore
Claude Opus 51552
Opus 4.61528
Claude Fable 51520
Opus 4.71513
GPT-5.6 Sol1511
Gemini 3.5 Flash1506
Opus 4.81504
Gemini 3.6 Flash1504
GPT-5.51503
GPT-5.41500
Muse Spark 1.11499
MiMo-V2.5-Pro1494
GLM-5.21493
Grok 4.51493
ERNIE 5.11487
Gemini 3.1 Pro Preview1487
Kimi K2.61485
GPT-5.6 Terra1485
Sonnet 51485
Sonnet 4.61484
GLM-5.11483
Gemini 3 Pro Preview1481
Qwen3.7-Plus1481
Kimi K2.51478
Opus 4.51478
GPT-5.6 Luna1478
Gemma 4 26B A4B1475
Gemma 4 31B1474
Gemini 3 Flash Preview1471
Qwen3.7 Max Preview1470
MiMo-V2-Pro1470
Muse Spark1469
GPT-5.21464
Inkling1462
GPT-5.11461
DeepSeek-V4-Pro1461
Qwen3.5 397B A17B1460
GLM-51457
Qwen3.6 Plus (2026-04-02)1456
MiMo-V2.51455
Loading Atlas data…