LMArena Elo (Chatbot Arena) — AI model leaderboard

AI models ranked by LMArena Elo (Chatbot Arena), an aggregated third-party benchmark from lmarena. Higher is better. Cross-referenced against our first-party meo scores and Effective Value (𝕍).

Ranking 37 models across the full field · as of 2025-07-18.

#	Model	Lab	LMArena Elo (Chatbot Arena)
1	Google: Gemini 2.5 Pro Preview 05-06	google	1446
2	OpenAI: o3	openai	1428
3	DeepSeek: R1 0528	deepseek	1424
4	Google: Gemini 2.5 Flash	google	1417
5	OpenAI: GPT-4.1	openai	1384
6	DeepSeek: DeepSeek V3 0324	deepseek	1382
7	DeepSeek: R1	deepseek	1373
8	Mistral: Mistral Medium 3	mistralai	1372
9	Anthropic: Claude Opus 4	anthropic	1368
10	Qwen: Qwen3 235B A22B	qwen	1367
11	MiniMax: MiniMax M1	minimax	1364
12	OpenAI: o4 Mini	openai	1363
13	Google: Gemma 3 27B	google	1359
14	OpenAI: o1	openai	1350
15	Qwen: Qwen3 32B	qwen	1344
16	Anthropic: Claude Sonnet 4	anthropic	1341
17	OpenAI: o3 Mini High	openai	1340
18	OpenAI: GPT-4.1 Mini	openai	1337
19	Google: Gemma 3 12B	google	1337
20	Cohere: Command A	cohere	1327
21	Qwen: Qwen3 30B A3B	qwen	1321
22	OpenAI: o3 Mini	openai	1318
23	Google: Gemma 3n 4B	google	1305
24	OpenAI: GPT-4o (2024-05-13)	openai	1300
25	Google: Gemma 3 4B	google	1291
26	OpenAI: GPT-4.1 Nano	openai	1287
27	OpenAI: GPT-4o-mini (2024-07-18)	openai	1287
28	OpenAI: GPT-4o (2024-08-06)	openai	1280
29	Mistral Large 2407	mistralai	1267
30	Anthropic: Claude 3.5 Haiku	anthropic	1254
31	Google: Gemma 2 27B	google	1235
32	Mistral: Mistral Small 3	mistralai	1233
33	Meta: Llama 3 70B Instruct	meta-llama	1222
34	Microsoft: Phi 4	microsoft	1221
35	Anthropic: Claude 3 Haiku	anthropic	1194
36	Meta: Llama 3 8B Instruct	meta-llama	1167
37	Google Gemini Pro Latest	~google	1126

LMArena / Chatbot Arena, via mathewhe/chatbot-arena-elo (HF, Apache-2.0)

← All rankings Methodology & 𝕍 →