Production Benchmarks

Real performance data from production traffic. Not synthetic benchmarks.

Live data3.2B tokens across 107 models from 2,755 developers

#	Model	Tokens	Speed	Uptime
1	deepseek-v4-flashby DeepSeek	1.1B	40tok/s	100%
2	qwen-3.6-plusby Alibaba	534.3M	41tok/s	100%
3	gemma-4-31bby Google	392.4M	7tok/s	100%
4	deepseek-v3by DeepSeek	181.0M	9tok/s	98%
5	gpt-oss-120bby OpenAI	125.5M	39tok/s	100%
6	minimax-m2.7by MiniMax	119.0M	25tok/s	100%
7	gemini-2.5-flashby Google	91.8M	11tok/s	99%
8	gemini-3-flashby Google	87.1M	82tok/s	99%
9	qwen-3-coderby Alibaba	79.6M	15tok/s	97%
10	deepseek-v4-proby DeepSeek	75.8M	15tok/s	98%
11	llama-3.3-70bby Meta	72.5M	7tok/s	99%
12	glm-4.7-flashby Zhipu AI	52.6M	65tok/s	100%
13	minimax-m2.5by MiniMax	48.1M	26tok/s	100%
14	qwen-3-32bby Alibaba	37.7M	17tok/s	99%
15	kimi-k2.5by Moonshot	35.2M	34tok/s	100%
16	deepseek-r1by DeepSeek	20.1M	19tok/s	98%
17	qwen3.7-flashby Alibaba	15.6M	141tok/s	96%
18	gemini-3.5-flashby Google	15.2M	18tok/s	84%
19	glm-4.5-airby Zhipu AI	14.8M	39tok/s	100%
20	glm-5.1by Zhipu AI	14.7M	16tok/s	100%
21	kimi-k2.6by Moonshot	12.4M	16tok/s	98%
22	glm-5.2by Zhipu AI	10.2M	49tok/s	100%
23	gemini-3.6-flashby Google	8.7M	—	98%
24	gpt-5.6-luna-proby OpenAI	8.2M	54tok/s	100%
25	qwen-3.8-maxby Alibaba	4.9M	42tok/s	99%
26	qwen-3.7-maxby Alibaba	4.4M	—	81%
27	grok-4.3by xAI	4.2M	38tok/s	100%
28	gemini-3-flash-audioby Google	3.8M	—	100%
29	gpt-5.6-lunaby OpenAI	3.5M	20tok/s	100%
30	kimi-k2.7-codeby Moonshot	3.0M	14tok/s	100%
31	qwen-3.7-plusby Alibaba	2.5M	—	99%
32	gpt-5.6-solby OpenAI	2.5M	44tok/s	100%
33	A claude-sonnet-5by Anthropic	1.8M	41tok/s	100%
34	kimi-k3by Moonshot	1.2M	19tok/s	100%
35	gemini-3.7-flashby Google	1.1M	80tok/s	81%
36	gpt-5.6-sol-proby OpenAI	869.2K	—	99%
37	grok-buildby xAI	868.9K	—	99%
38	A claude-sonnet-4-6by Anthropic	703.3K	6tok/s	100%
39	whisper-v3-turboby OpenAI	677.3K	—	100%
40	A claude-opus-5by Anthropic	652.9K	37tok/s	98%
41	gpt-5.6-terraby OpenAI	645.6K	—	100%
42	sonarby Perplexity	515.7K	—	100%
43	nemotron-3-ultra-550bby NVIDIA	452.6K	30tok/s	98%
44	minimax-m3by MiniMax	401.2K	7tok/s	98%
45	grok-4.20-multi-agentby xAI	350.1K	321tok/s	100%
46	kling-3-proby Kuaishou	330.8K	—	100%
47	gpt-5.6-terra-proby OpenAI	274.5K	51tok/s	100%
48	seedance-2-proby ByteDance	253.6K	—	100%
49	gemini-3.5-flash-liteby Google	219.2K	43tok/s	99%
50	grok-4.5by xAI	184.6K	18tok/s	99%
51	kling-3-pro-audioby Kuaishou	160.8K	—	100%
52	kling-2.5-proby Kuaishou	149.4K	—	100%
53	seedance-2-fastby ByteDance	138.7K	—	100%
54	A claude-haiku-4-5by Anthropic	112.8K	—	100%
55	flux-1.1-ultraby Black Forest Labs	112.4K	—	100%
56	ideogram-v3by Ideogram	111.8K	—	100%
57	flux-kontext-proby Black Forest Labs	104.1K	—	100%
58	A claude-opus-4-7by Anthropic	88.2K	3tok/s	96%
59	grok-4.6by xAI	81.5K	—	85%
60	llama-4-maverickby Meta	79.1K	—	100%
61	muse-spark-1.1by Meta	59.7K	62tok/s	100%
62	qwen3-embedding-8bby Alibaba	55.8K	—	100%
63	grok-4.20by xAI	53.2K	—	100%
64	A claude-opus-5-fastby Anthropic	30.6K	—	100%
65	gpt-4o-mini-transcribe-2025-12-15by OpenAI	18.9K	—	100%
66	B bge-m3by BAAI	12.8K	—	100%
67	qwen3-reranker-8bby Alibaba	11.7K	—	100%
68	sonar-proby Perplexity	11.5K	—	91%
69	gemini-3.1-proby Google	8.0K	18tok/s	100%
70	M multilingual-e5-largeby Microsoft	6.8K	—	88%
71	embeddinggemma-300mby Google	6.0K	—	100%
72	gte-baseby Alibaba	5.4K	—	90%
73	step-3.7-flashby StepFun	5.1K	—	96%
74	B bge-large-enby BAAI	4.9K	—	100%
75	B bge-base-enby BAAI	4.9K	—	100%
76	qwen3-embedding-4bby Alibaba	4.8K	—	100%
77	qwen3-embedding-0.6bby Alibaba	4.8K	—	100%
78	N hermes-3-405bby Nous Research	4.3K	—	31%
79	S all-minilm-l6by Sentence Transformers	4.0K	—	83%
80	recraft-v3by Recraft	2.9K	—	100%
81	A claude-fable-5by Anthropic	2.6K	20tok/s	80%
82	S all-minilm-l12by Sentence Transformers	2.6K	—	89%
83	N hermes-3-70bby Nous Research	392	—	100%
84	eleven-flash-v2-5by ElevenLabs	112	—	100%
85	glm-5.3by Zhipu AI	111	16tok/s	67%
86	hailuo-02-768pby MiniMax	56	—	100%
87	hailuo-02-1080pby MiniMax	56	—	100%
88	minimax-musicby MiniMax	56	—	100%
89	eleven-multilingual-v2by ElevenLabs	56	—	91%
90	minimax-speech-hdby MiniMax	56	—	100%
91	hailuo-02-512pby MiniMax	56	—	100%
92	minimax-speech-turboby MiniMax	56	—	100%
93	minimax-voice-cloneby MiniMax	56	—	100%
94	eleven-turbo-v2-5by ElevenLabs	56	—	100%
95	minimax-voice-designby MiniMax	56	—	100%
96	recraft-v4by Recraft	56	—	100%
97	recraft-v4-proby Recraft	56	—	100%
98	recraft-v4-vectorby Recraft	56	—	100%
99	recraft-v4-vector-proby Recraft	56	—	100%
100	elevenlabs-musicby ElevenLabs	56	—	100%
101	elevenlabs-sfxby ElevenLabs	56	—	100%
102	minimax-image-01by MiniMax	56	—	100%
103	flux-2-proby Black Forest Labs	56	—	100%
104	minimax-music-proby MiniMax	56	—	100%
105	glm-5.3-flashby Zhipu AI	0	—	0%
106	eleven-v3by ElevenLabs	0	—	99%
107	qwen3.8-flashby Alibaba	0	—	0%

Measured from real API traffic. Updated every 5 minutes. Read docs →