Ranked by real usage: token volume routed per model on OpenRouter over the last 7 days · ● Live, updated 2026-08-18
As of 2026-08-18, the most used AI model tracked here is DeepSeek V4 Flash 0731 (DeepSeek), followed by Hy3 and GPT-5.6 Luna (batch). Ranked by tokens routed per model on OpenRouter over the trailing 7 days; refreshed automatically.
| # | Model Name | Tokens/wk |
|---|---|---|
| 1 | DeepSeek V4 Flash 0731 |
16.07T
|
| 2 | Hy3 Chinese tech giant behind the Hunyuan / Hy models |
9.7T
|
| 3 | GPT-5.6 Luna (batch) US lab behind ChatGPT and the GPT series |
5.57T
|
| 4 | MiMo-V2.5 Chinese electronics giant, maker of the open MiMo models |
4.99T
|
| 5 | GLM 5.2 Chinese lab (Zhipu AI) behind the open GLM models |
4.41T
|
| 6 | DeepSeek V4 Pro 0813 |
3.45T
|
| 7 | Gemini 3.6 Flash (batch) Google DeepMind, maker of the Gemini models |
2.77T
|
| 8 | Claude Opus 5 (batch) US AI-safety lab, maker of Claude |
2.7T
|
| 9 | Nemotron 3 Ultra (free) NVIDIA's model line (Nemotron) |
2.44T
|
| 10 | Laguna S 2.1 (free) From poolside |
1.68T
|
| 11 | MiniMax M3 (batch) Chinese AI lab (Shanghai) building open-weight models |
1.55T
|
| 12 | Kimi K3 Chinese lab (Moonshot AI) behind the Kimi models |
1.37T
|
| 13 | Claude Sonnet 5 (batch) US AI-safety lab, maker of Claude |
1.08T
|
| 14 | GPT-5.6 Terra (batch) US lab behind ChatGPT and the GPT series |
948B
|
| 15 | Step 3.7 Flash From stepfun |
922B
|
| 16 | Gemini 3 Flash Preview (batch) Google DeepMind, maker of the Gemini models |
816B
|
| 17 | Nemotron 3.5 Lightning (free) NVIDIA's model line (Nemotron) |
797B
|
| 18 | GPT-5.6 Sol (batch) US lab behind ChatGPT and the GPT series |
747B
|
| 19 | Gemini 2.5 Flash Lite (batch) Google DeepMind, maker of the Gemini models |
717B
|
| 20 | Claude Sonnet 4.6 (batch) US AI-safety lab, maker of Claude |
706B
|
| 21 | GPT-5.6 Luna Pro (batch) US lab behind ChatGPT and the GPT series |
613B
|
| 22 | Gemini 3.7 Flash (batch) Google DeepMind, maker of the Gemini models |
603B
|
| 23 | Claude Opus 4.8 (batch) US AI-safety lab, maker of Claude |
580B
|
| 24 | Claude Opus 4.7 (batch) US AI-safety lab, maker of Claude |
524B
|
| 25 | MiMo-V2.5-Pro Chinese electronics giant, maker of the open MiMo models |
489B
|
| 26 | Gemma 4 31B (free) Google DeepMind, maker of the Gemini models |
468B
|
| 27 | gpt-oss-120b US lab behind ChatGPT and the GPT series |
458B
|
| 28 | Gemini 2.5 Flash (batch) Google DeepMind, maker of the Gemini models |
449B
|
| 29 | Gemini 3.1 Flash Lite (batch) Google DeepMind, maker of the Gemini models |
420B
|
| 30 | DeepSeek V3.2 |
411B
|
| 31 | Gemma 4 26B A4B (free) Google DeepMind, maker of the Gemini models |
378B
|
| 32 | Solar Pro 4 From upstage |
377B
|
| 33 | Nemotron 3 Super (free) NVIDIA's model line (Nemotron) |
362B
|
| 34 | Grok 4.6 From SpaceXAI |
354B
|
| 35 | GPT-5.5 (batch) US lab behind ChatGPT and the GPT series |
300B
|
| 36 | Claude Fable 5 (batch) US AI-safety lab, maker of Claude |
283B
|
| 37 | Qwen3.8 Max Alibaba lab behind the open Qwen models |
276B
|
| 38 | Grok 4.5 From SpaceXAI |
249B
|
| 39 | Claude Haiku 4.5 (batch) US AI-safety lab, maker of Claude |
237B
|
| 40 | North Mini Code (free) Enterprise-focused AI lab (Command models) |
218B
|
| 41 | GPT-4o-mini (batch) US lab behind ChatGPT and the GPT series |
205B
|
| 42 | GPT-5.4 (batch) US lab behind ChatGPT and the GPT series |
201B
|
| 43 | MiniMax M2.7 Chinese AI lab (Shanghai) building open-weight models |
197B
|
| 44 | Claude Opus 4.6 (batch) US AI-safety lab, maker of Claude |
193B
|
| 45 | Mistral Nemo |
183B
|
| 46 | Gemini 3.1 Pro Preview (batch) Google DeepMind, maker of the Gemini models |
179B
|
| 47 | Gemini 3.5 Flash (batch) Google DeepMind, maker of the Gemini models |
174B
|
| 48 | Gemini 3.5 Flash Lite (batch) Google DeepMind, maker of the Gemini models |
174B
|
| 49 | Ling-2.6-flash From inclusionai |
164B
|
| 50 | Laguna XS 2.1 (free) From poolside |
163B
|
| 51 | GPT-5 Mini (batch) US lab behind ChatGPT and the GPT series |
159B
|
| 52 | Qwen3.7 Flash Alibaba lab behind the open Qwen models |
158B
|
| 53 | gpt-oss-20b (free) US lab behind ChatGPT and the GPT series |
155B
|
| 54 | Qwen3.6 35B A3B Alibaba lab behind the open Qwen models |
154B
|
| 55 | Kimi K2.6 Chinese lab (Moonshot AI) behind the Kimi models |
149B
|
| 56 | Qwen3.7 Max Alibaba lab behind the open Qwen models |
140B
|
| 57 | Kimi K2.7 Code (batch) Chinese lab (Moonshot AI) behind the Kimi models |
112B
|
| 58 | GPT-5.4 Nano (batch) US lab behind ChatGPT and the GPT series |
109B
|
| 59 | GPT-5.4 Mini (batch) US lab behind ChatGPT and the GPT series |
104B
|
| 60 | Ling-3.0-flash From inclusionai |
96B
|
| 61 | Gemini 3.1 Flash Lite Preview Google DeepMind, maker of the Gemini models |
96B
|
| 62 | Muse Spark 1.2 Meta AI, maker of the open Llama models |
95B
|
| 63 | Qwen3.7 Plus Alibaba lab behind the open Qwen models |
88B
|
| 64 | Kimi K2.5 Chinese lab (Moonshot AI) behind the Kimi models |
84B
|
| 65 | GPT-4.1 Mini (batch) US lab behind ChatGPT and the GPT series |
80B
|
| 66 | Claude Sonnet 4.5 (batch) US AI-safety lab, maker of Claude |
75B
|
| 67 | GLM 5.1 Chinese lab (Zhipu AI) behind the open GLM models |
70B
|
| 68 | Qwen3 235B A22B Instruct 2507 Alibaba lab behind the open Qwen models |
68B
|
| 69 | Llama 3.1 8B Instruct Meta AI, maker of the open Llama models |
67B
|
| 70 | Grok 4.3 From SpaceXAI |
60B
|
| 71 | Gemma 3 27B Google DeepMind, maker of the Gemini models |
59B
|
| 72 | GLM 4.7 Chinese lab (Zhipu AI) behind the open GLM models |
58B
|
| 73 | Gemini 2.5 Pro (batch) Google DeepMind, maker of the Gemini models |
56B
|
| 74 | Nova Micro 1.0 From Amazon |
55B
|
| 75 | Qwen3.5-Flash Alibaba lab behind the open Qwen models |
54B
|
| 76 | GPT-5.6 Sol Pro (batch) US lab behind ChatGPT and the GPT series |
52B
|
| 77 | Nemotron 3 Nano 30B A3B (free) NVIDIA's model line (Nemotron) |
52B
|
| 78 | Nex-N2-Mini From nex-agi |
51B
|
| 79 | GPT-4.1 (batch) US lab behind ChatGPT and the GPT series |
50B
|
| 80 | GLM 5 Chinese lab (Zhipu AI) behind the open GLM models |
49B
|
| 81 | GPT-5 Nano (batch) US lab behind ChatGPT and the GPT series |
46B
|
| 82 | Llama 4 Scout Meta AI, maker of the open Llama models |
45B
|
| 83 | GPT-5.6 Terra Pro (batch) US lab behind ChatGPT and the GPT series |
44B
|
| 84 | DeepSeek V3.1 |
43B
|
| 85 | MiniMax M2.5 Chinese AI lab (Shanghai) building open-weight models |
42B
|
| 86 | Qwen3.5 397B A17B Alibaba lab behind the open Qwen models |
39B
|
| 87 | Muse Glimmer 30B Meta AI, maker of the open Llama models |
39B
|
| 88 | GPT-5.3-Codex US lab behind ChatGPT and the GPT series |
36B
|
| 89 | DeepSeek V3 0324 |
36B
|
| 90 | Ling 3.0 Tiny From inclusionai |
35B
|
| 91 | Grok 4.20 From SpaceXAI |
35B
|
| 92 | GPT-4.1 Nano (batch) US lab behind ChatGPT and the GPT series |
33B
|
| 93 | Dots3-Note Preview (free) |
32B
|
| 94 | Llama 3.3 70B Instruct Meta AI, maker of the open Llama models |
32B
|
| 95 | Nemotron 3 Nano Omni (free) NVIDIA's model line (Nemotron) |
32B
|
| 96 | Qwen3 VL 32B Instruct Alibaba lab behind the open Qwen models |
31B
|
| 97 | Qwen3 32B Alibaba lab behind the open Qwen models |
30B
|
| 98 | Gemma 3 12B Google DeepMind, maker of the Gemini models |
29B
|
| 99 | GLM 5V Turbo Chinese lab (Zhipu AI) behind the open GLM models |
27B
|
| 100 | Claude Opus 5 (Fast) US AI-safety lab, maker of Claude |
27B
|
Recently listed on OpenRouter, not yet in the ranking above — too new to have meaningful usage yet.
| Model Name | Tokens/wk | Listed |
|---|---|---|
| Qwen3.8 27B | — | 2026-08-14 |
| Dots3-Note Preview (free) | — | 2026-08-14 |
| Gemini 3.7 Flash (batch) | — | 2026-08-13 |
| voyage-code-4 | — | 2026-08-13 |
| Qwen3 Reranker 8B | — | 2026-08-13 |
Ranked by real usage: the total tokens (prompt + completion) routed to each model on OpenRouter over the trailing 7 days. This reflects developer/API demand on the OpenRouter marketplace. It does not include first-party app traffic (ChatGPT, the Gemini app, claude.ai), so consumer chat flagships are under-counted relative to their true reach. Names, pricing and context windows are live from OpenRouter's own catalog and validated against it before display. Refreshed automatically; no hand-maintained list.