Large Language Models
docAnalyzer keeps pace with releases from Alibaba, Anthropic, DeepSeek, Google, Minimax, Moonshot AI, OpenAI, Xiaomi, and xAI, adding new models within days so you can pick the right fit for every workflow. We surface 28 models across 9 creators, each benchmarked for quality, speed, and responsiveness.
Refreshed 2026-05-23
60.2 top quality · GPT-5.5
330 tok/s top speed · GPT OSS 120b
860 ms top responsiveness · DeepSeek V4 Flash
| Model | |||
|---|---|---|---|
 GPT-5.5 OpenAI | 60.2 | 65 tok/s | 73.84 s |
 Claude Opus 4.7 Anthropic | 57.3 | 51 tok/s | 13.80 s |
 Gemini 3.1 Pro Preview Google | 57.2 | 131 tok/s | 23.69 s |
| GPT‑5.4 OpenAI | 56.8 | 95 tok/s | 202.43 s |
| Gemini 3.5 Flash Google | 55.3 | 225 tok/s | 13.38 s |
 Kimi K2.6 Moonshot AI | 53.9 | 56 tok/s | 1.06 s |
 Grok 4.3 xAI | 53.2 | 117 tok/s | 25.98 s |
| Claude Sonnet 4.6 Anthropic | 51.7 | 75 tok/s | 51.62 s |
 DeepSeek V4 Pro DeepSeek | 51.5 | 32 tok/s | 1.12 s |
 Qwen3.6 Plus Alibaba | 50.0 | 52 tok/s | 1.69 s |
 MiniMax M2.7 Minimax | 49.6 | 53 tok/s | 1.31 s |
 MiMo V2.5 Xiaomi | 49.0 | 95 tok/s | 2.57 s |
| GPT-5.4 mini OpenAI | 48.9 | 164 tok/s | 10.52 s |
| Grok 4.20 Beta (non reasoning) xAI | 48.5 | 98 tok/s | 37.46 s |
| Grok 4.20 Beta (reasoning) xAI | 48.5 | 98 tok/s | 37.46 s |
| Kimi K2.5 Moonshot AI | 46.8 | 36 tok/s | 1.33 s |
| DeepSeek V4 Flash DeepSeek | 46.5 | 108 tok/s | 860 ms |
| Gemini 3 Flash Preview Google | 46.4 | 187 tok/s | 6.13 s |
| Qwen3.5-397B-A17B Alibaba | 45.0 | 52 tok/s | 1.80 s |
| GPT-5.4 nano OpenAI | 44.0 | 156 tok/s | 2.43 s |
| Qwen3.5-27B Alibaba | 42.1 | 91 tok/s | 1.44 s |
| MiMo V2 Flash Xiaomi | 41.5 | 133 tok/s | 1.53 s |
| GPT-5 mini OpenAI | 41.2 | 90 tok/s | 77.98 s |
| Claude Haiku 4.5 Anthropic | 37.1 | 133 tok/s | 12.28 s |
| Gemini 3.1 Flash Lite Preview Google | 33.5 | 327 tok/s | 5.25 s |
| GPT OSS 120b OpenAI | 33.3 | 330 tok/s | 520 ms |
| GPT-5 nano OpenAI | 26.8 | 156 tok/s | 97.95 s |
| Qwen 3 235B Alibaba | 25.0 | 67 tok/s | 1.19 s |
* Latency uses a logarithmic scale: shorter time means a longer bar.