Benchmark

Arena ELO

Chatbot Arena Elo Rating

Crowd-sourced pairwise preference ranking from live human evaluations.

Models Tested

Gemini 2.5 Pro

Top Model

1398

Top Score

1600 ELO

Max Possible

#	Model	Arena ELO Score	Provider	Composite
1	Gemini 2.5 Pro Google	1398	Google	97.5
2	Claude Opus 4 Anthropic	1385	Anthropic	97.8
3	o3 OpenAI	1380	OpenAI	97.1
4	DeepSeek R1 DeepSeek	1358	DeepSeek	92.3
5	o4-mini OpenAI	1345	OpenAI	91.5
6	Claude Sonnet 4 Anthropic	1340	Anthropic	90.3
7	Grok 3 xAI	1330	xAI	88.8
8	Llama 4 Maverick Meta	1325	Meta	87.5
9	DeepSeek V3 DeepSeek	1318	DeepSeek	86.1
10	Gemini 2.5 Flash Google	1310	Google	86.7
11	Qwen 3 235B Alibaba	1305	Alibaba	84.5
12	Gemini 2.0 Flash Google	1290	Google	80.5
13	GPT-4o OpenAI	1287	OpenAI	89.2
14	Claude 3.5 Sonnet Anthropic	1272	Anthropic	88.4
15	GPT-4o Mini OpenAI	1271	OpenAI	75.8
16	Llama 3.1 405B Meta	1265	Meta	82.1
17	Llama 3.3 70B Meta	1258	Meta	79.8
18	Claude 3.5 Haiku Anthropic	1248	Anthropic	79.2
19	Mistral Large Mistral AI	1245	Mistral AI	76.5
20	Grok 2 xAI	1240	xAI	74.8
21	Llama 3.1 70B Meta	1235	Meta	74.3
22	Qwen 2.5 72B Alibaba	1230	Alibaba	76
23	Mixtral 8x22B Mistral AI	1195	Mistral AI	65.2
24	Command R+ Cohere	1185	Cohere	60.8
25	Phi-4 Microsoft	1150	Microsoft	64.5

Source: LMSYS, 2024