AI Models Leaderboard

34 current models ranked by benchmark score. Click a column header to re-sort.

#	Model	Provider					Context	Pricing
1	Grok 3	xAI	93.3	n/a	97.7	84.6	131K	Freemium
2	Claude Opus 4.6featured	Anthropic	92.0	94.5	97.1	79.9	200K	Paid
3	Gemini 2.5 Profeatured	Google	91.0	n/a	97.0	84.0	1.05M	Paid
4	DeepSeek R1featured	DeepSeek	90.8	n/a	97.3	71.5	66K	Open
5	Claude Sonnet 4.6featured	Anthropic	90.1	93.5	93.7	70.0	200K	Freemium
6	GPT-4ofeatured	OpenAI	88.7	90.2	76.6	53.6	128K	Freemium
7	Llama 3.1 405B	Meta	88.6	n/a	73.5	n/a	131K	Open
8	DeepSeek V3	DeepSeek	88.5	91.6	90.2	n/a	66K	Open
9	Grok 2	xAI	87.5	n/a	76.1	n/a	131K	Freemium
10	Qwen2.5-72B-Instruct	Alibaba	86.1	86.6	83.1	n/a	131K	Open
11	Llama 3.3 70B	Meta	86.0	n/a	77.0	n/a	131K	Open
12	Phi-4	Microsoft	84.8	82.6	80.4	56.1	16K	Open
13	Mistral Large 2	Mistral AI	84.0	92.0	69.7	n/a	131K	Paid
14	GPT-4o mini	OpenAI	82.0	87.2	n/a	n/a	128K	Freemium
15	Mistral Small 3	Mistral AI	81.0	n/a	n/a	n/a	33K	Open
16	Command R+	Cohere	75.7	n/a	n/a	n/a	128K	Paid
17	Llama 3.2 11B Vision	Meta	73.0	n/a	n/a	n/a	131K	Open
18	Claude Fable 5.2featured	Anthropic	n/a	n/a	n/a	n/a	1M	Paid
19	Claude Opus 5.5featured	Anthropic	n/a	n/a	n/a	n/a	1M	Paid
20	Gemini 4 Carbonfeatured	Google	n/a	n/a	n/a	n/a	n/a	Paid
21	GPT-6.1 Solfeatured	OpenAI	n/a	n/a	n/a	n/a	1.05M	Paid
22	GPT-6 Astrafeatured	OpenAI	n/a	n/a	n/a	n/a	1.05M	Paid
23	Claude Haiku 5.5	Anthropic	n/a	n/a	n/a	n/a	1M	Paid
24	Claude Sonnet 5.5	Anthropic	n/a	n/a	n/a	n/a	1M	Paid
25	Codestral	Mistral AI	n/a	91.1	n/a	n/a	33K	Freemium
26	DeepSeek V4.1 Flash	DeepSeek	n/a	n/a	n/a	n/a	1M	Open
27	DeepSeek V4 Pro	DeepSeek	n/a	n/a	n/a	n/a	1M	Paid
28	Gemini 3.8 Flash	Google	n/a	n/a	n/a	n/a	n/a	Paid
29	Gemini 4 Argon	Google	n/a	n/a	n/a	n/a	n/a	Paid
30	GPT-6.1 Terra	OpenAI	n/a	n/a	n/a	n/a	1.05M	Paid
31	GPT-6 Luna	OpenAI	n/a	n/a	n/a	n/a	1.05M	Paid
32	Grok 4.7	xAI	n/a	n/a	n/a	n/a	500K	Paid
33	Mistral Large 4	Mistral AI	n/a	n/a	n/a	n/a	n/a	Open
34	Mistral Small 4	Mistral AI	n/a	n/a	n/a	n/a	n/a	Open

Benchmark scores are sourced from official provider publications and independent evaluations. Scores reflect the model version and evaluation methodology at the time of measurement, so direct comparisons across providers should be treated as approximate.