Atlas

Benchmarks

← All benchmarks

LMArena Text Arena Coding Elo

Code · 2024-04-14

LMArena text-arena coding leaderboard Elo for pairwise human preference votes on coding-related prompts. Higher Elo indicates stronger human-preference performance.

Top models (higher is better)

ModelScore
Gemini 4 Argon1552
Claude Fable 51551
Opus 4.71550
Claude Opus 5.51548
Opus 4.61547
Muse Wonder1545
GPT-6 Astra1543
Muse Spark 1.21541
Kimi K31539
MiMo-V2.6-Pro1538
Claude Fable 5.11537
GPT-6.1 Sol1537
Muse Spark 1.31536
Claude Opus 51534
GPT-5.6 Sol1534
Opus 4.81533
Muse Spark 1.11533
Muse Spark1529
Sonnet 4.61528
Gemini 3.8 Flash1528
Claude Sonnet 5.51528
GLM-5.3 Flash1525
Qwen3.7 Max Preview1524
Qwen3.8-Max1524
DeepSeek V4.1 Flash1523
Opus 4.51523
Gemini 3.1 Pro Preview1522
GLM-5.31521
GPT-5.41521
MiMo-V2.5-Pro1521
Gemini 3.7 Flash1520
Sonnet 51520
GPT-6 Sol1519
GPT-5.51519
Gemini 3.6 Flash1518
GPT-5.6 Terra1518
Gemini 3 Pro Preview1517
Kimi K2.61516
GLM-5.21514
Grok 4.51514
Loading Atlas data…