⌕Is ___ SOTA?
benchmarks / lmarena-text-hard

LMArena Text (style-controlled, hard prompts)

Arena score. Status active. Contamination risk low. Maintainer LMArena. Source.

ModelScoreDateKind
ChatGPT 4O Latest1455.14979790926622026-10-13snapshot
Claude 3.5 Sonnet1399.21589209177452026-10-13snapshot
GPT-4O1252.7707120561252025-02-06snapshot
Llama 3.1 405B Instruct1248.78136764503372024-09-04snapshot
Grok 21239.81545028378382024-12-18snapshot
Gemini 1.5 Pro Api 05141238.18516795509842024-10-15snapshot
Gemini Advanced 05141236.3430271013782024-10-23snapshot
Claude 3 Opus1234.57288009991822024-10-07snapshot
Mistral Large 24071230.89340673682322024-09-27snapshot
GPT-4 1106 Preview1227.57695735588782024-09-15snapshot
GPT-4 Turbo1225.17340285019282024-09-04snapshot
Llama 3.1 405B Instruct Bf161249.89065608424582025-02-03snapshot
Llama 3.1 405B Instruct Fp81248.1636874786682025-01-22snapshot
DeepSeek V2.51227.65280388590052024-09-15snapshot
O1 Preview1396.40688259058722026-10-13snapshot
O1 Mini1279.70317328864532025-03-11snapshot
Gemini 1.5 Pro 0021263.74177336398682025-02-17snapshot
YI Lightning1252.68118166826532025-02-05snapshot
Gemini 1.5 Pro 0011237.84574089683522024-11-21snapshot
Athene V2 Chat1245.26076636048362024-12-22snapshot
Claude 3.5 Haiku1253.29873606301732025-01-05snapshot
Qwen2 5 Plus 11271247.75482624767642024-12-22snapshot
O11418.00197067598832026-10-13snapshot
DeepSeek V31258.14737156053052025-02-11snapshot
Step 2 16k Exp 2024121248.37511435540732025-01-28snapshot
DeepSeek R11418.69485451125662026-10-13snapshot
Qwen2 51385.09839377244732026-10-13snapshot
Gemini 2.0 Flash 0011287.2917297923952025-03-25snapshot
O3 Mini1302.96081425902962025-04-16snapshot
Gemini 2.0 Flash Lite Preview 02.051267.33031284535012025-02-27snapshot