Model Leaderboard

Which AI model is actually best — real benchmarks, real usage, synced daily. Not vibes.

Free preview

Every table below is open to everyone. Premium unlocks BREAF's own written take on why a model earned its spot.

See Premium

10 newest models — best available score

Claude Opus 5

40.5

GPT-5.6 Sol

39.7

GLM-5.3

39.4

Claude Fable 5

38.6

Kimi K3

38.0

DeepSeek-V4-Pro-0813

35.9

GPT-5.6 Terra

35.4

Claude Mythos Preview

35.3

GPT-5.5 Pro

35.1

Grok 4.6

34.5

#	Model	BREAF Score	MMLU-Pro	GPQA	SWE-Bench	AIME	Agent	HLE	ARC-AGI v2	MMMLU	BrowseComp	OSWorld	SimpleQA	Price in/M	Price out/M	Speed	License
1	Claude Opus 5Anthropic	40.5	—	—	—	—	—	65%	—	—	91%	—	—	$5.00	$25.00	—	Proprietary
2	GPT-5.6 SolOpenAI	39.7	—	95%	—	—	53%	—	—	—	90%	—	—	$5.00	$30.00	—	Proprietary
3	GLM-5.3Zhipu AI	39.4	—	—	—	—	28%	63%	—	—	—	—	—	$1.40	$4.40	—	Unknown
4	Claude Fable 5Anthropic	38.6	—	—	95%	—	—	65%	—	—	—	—	—	$10.00	$50.00	—	Proprietary
5	Kimi K3Moonshot AI	38.0	—	94%	—	—	—	56%	—	—	91%	—	—	$3.00	$15.00	—	Kimi K3 License
6	DeepSeek-V4-Pro-0813DeepSeek	35.9	—	—	—	—	26%	60%	—	—	—	—	—	$0.44	$0.87	—	MIT
7	GPT-5.6 TerraOpenAI	35.4	—	93%	—	—	50%	—	—	—	88%	—	—	$2.00	$12.00	—	Proprietary
8	Claude Mythos PreviewAnthropic	35.3	—	95%	94%	—	—	65%	—	93%	87%	—	—	—	—	—	Proprietary
9	GPT-5.5 ProOpenAI	35.1	—	—	—	—	—	57%	—	—	90%	—	—	—	—	—	Proprietary
10	Grok 4.6xAI	34.5	—	—	—	—	—	—	—	—	—	—	—	$2.00	$6.00	—	Proprietary
11	Gemini 3.7 FlashGoogle	34.4	—	—	—	—	26%	—	—	—	—	—	—	$0.75	$3.75	—	Proprietary
12	Qwen3.8 MaxAlibaba Cloud / Qwen Team	33.7	—	93%	—	—	52%	44%	—	—	—	—	—	$1.65	$4.95	—	Qwen3.8-Max License
13	DeepSeek-V4-Flash-Vision-ExpDeepSeek	32.6	—	—	—	—	27%	—	—	—	—	—	—	$0.22	$0.66	—	Unknown
14	GLM-5.3-FlashZhipu AI	32.2	—	—	—	—	26%	55%	—	—	—	—	—	$0.15	$0.50	—	MIT
15	Claude Opus 4.8Anthropic	31.5	—	94%	89%	—	—	58%	—	—	84%	—	—	$5.00	$25.00	42 tok/s	Proprietary
16	Gemini 3.5 Flash CyberGoogle	31.0	—	—	—	—	—	—	—	—	—	—	—	—	—	—	Proprietary
17	GPT-5.6 LunaOpenAI	30.6	—	92%	—	—	50%	—	—	—	83%	—	—	$0.20	$1.20	—	Proprietary
18	GPT-5.5OpenAI	29.1	—	94%	—	—	—	52%	85%	—	84%	—	—	$5.00	$30.00	—	Proprietary
19	Qwen3.8-Flash-NextAlibaba Cloud / Qwen Team	29.0	—	92%	—	—	51%	36%	—	—	—	—	—	—	—	—	Qwen Community License 1.0
20	Muse Spark 1.2Meta	28.8	—	—	—	—	—	—	—	—	—	—	—	$0.10	$0.20	—	Proprietary
21	Grok 4.5xAI	28.4	—	93%	—	—	—	—	—	—	—	—	—	$2.00	$6.00	80 tok/s	Proprietary
22	DeepSeek-V4-Flash-0731DeepSeek	28.2	—	—	—	—	25%	—	—	—	—	—	—	$0.09	$0.18	—	MIT
23	Muse Spark 1.1Meta	27.9	—	—	—	—	—	62%	—	—	—	—	—	$1.25	$4.25	—	Proprietary
24	Claude Sonnet 5Anthropic	27.8	—	—	85%	—	—	57%	—	—	85%	—	—	$2.00	$10.00	42 tok/s	Proprietary
25	Seed 2.1 ProByteDance	26.4	—	—	—	—	41%	56%	—	—	86%	79%	—	—	—	—	Proprietary
26	GLM-5.2Zhipu AI	24.4	—	91%	—	—	—	55%	—	—	—	—	—	$0.95	$3.00	—	MIT
27	Qwen3.8-27BAlibaba Cloud / Qwen Team	24.3	—	89%	—	—	43%	31%	—	—	—	—	—	—	—	—	Apache 2.0
28	Gemini 3.6 FlashGoogle	23.8	—	—	—	—	—	—	—	—	—	—	—	$1.50	$7.50	—	Proprietary
29	GLM-5V-TurboZhipu AI	23.6	—	—	—	—	—	—	—	—	—	62%	—	—	—	—	Proprietary
30	Claude Opus 4.7Anthropic	23.5	—	94%	88%	—	—	55%	—	92%	79%	—	—	$5.00	$25.00	42 tok/s	Proprietary
31	Qwen3.7 MaxAlibaba Cloud / Qwen Team	22.8	90%	92%	80%	—	—	41%	—	90%	—	—	—	$1.25	$3.75	—	Proprietary
32	Sakana NamazuSakana AI	22.7	90%	—	—	—	—	—	—	—	—	—	—	$0.95	$4.00	—	Proprietary
33	Kimi K2.6Moonshot AI	22.6	—	91%	80%	—	—	36%	—	—	86%	—	—	$0.75	$3.50	—	Modified MIT License
34	GPT-5.4OpenAI	22.4	—	93%	—	—	—	40%	73%	—	83%	—	—	$2.50	$15.00	50 tok/s	Proprietary
35	Qwen3.7-PlusAlibaba Cloud / Qwen Team	22.0	89%	90%	78%	—	—	35%	—	89%	—	—	—	—	—	—	Proprietary
36	Seed 2.1 TurboByteDance	21.9	—	—	—	—	—	55%	—	—	85%	76%	—	—	—	—	Proprietary
37	Muse SparkMeta	21.7	—	90%	77%	—	—	58%	43%	—	—	—	—	—	—	—	Proprietary
38	Gemini 3 ProGoogle	21.4	—	92%	76%	100%	—	46%	31%	92%	—	—	72%	—	—	—	Proprietary
39	GPT-5.2OpenAI	21.3	—	92%	80%	100%	—	35%	53%	90%	66%	—	—	$1.75	$14.00	100 tok/s	Proprietary
40	Claude 3.5 SonnetAnthropic	20.9	78%	67%	49%	—	—	—	—	—	—	—	—	—	—	—	Proprietary
41	Claude Opus 4.6Anthropic	20.5	—	91%	81%	100%	—	53%	69%	91%	84%	73%	—	$5.00	$25.00	42 tok/s	Proprietary
42	Hy3Tencent	20.4	—	90%	78%	—	—	—	—	—	84%	—	—	—	—	—	Apache 2.0
43	DeepSeek-V4-Flash-MaxDeepSeek	20.2	86%	88%	79%	—	—	45%	—	—	73%	—	34%	$0.14	$0.28	—	MIT
44	Gemini 3.5 FlashGoogle	19.9	—	—	—	—	—	40%	72%	—	—	—	—	$1.50	$9.00	—	Proprietary
45	GLM-5.1Zhipu AI	19.6	—	86%	—	—	—	52%	—	—	79%	—	—	$1.40	$4.40	—	MIT
46	GPT-5 MediumOpenAI	18.4	—	88%	—	89%	—	—	—	—	—	—	—	—	—	—	Proprietary
47	Hermes 3 70BNous Research	18.3	47%	66%	—	—	—	—	—	—	—	—	—	—	—	—	Apache 2.0
48	Qwen3.5-122B-A10BAlibaba Cloud / Qwen Team	18.3	87%	87%	72%	—	—	48%	—	87%	64%	—	—	—	—	—	Apache 2.0
49	GPT-5OpenAI	17.8	—	86%	75%	95%	—	25%	—	—	55%	—	—	—	—	—	Proprietary
50	Kimi K2.5Moonshot AI	17.8	87%	88%	77%	96%	—	50%	—	—	75%	—	—	—	—	—	MIT

Synced Aug 28 · Text/Image/Video: Source: llm-stats.com. Most Used/Free Right Now: Source: OpenRouter (openrouter.ai/rankings), licensed CC BY 4.0.