Real-world usage × price
What developers actually run — and what it costs
Weekly token volume across 97 models, from OpenRouter's public rankings (CC BY 4.0), joined with our verified pricing. Usage ≠ quality: it shows what the market has already settled on.
Data through 2026-10-21 · refreshed daily
Weekly token volume
160.7T
across 97 ranked models
Est. weekly spend
$172.8M
at blended 3:1 in:out pricing
📈 Rising fast this week
Qwen: Qwen3.6 35B A3B +366.9%Meta: Muse Wonder +203.8%OpenAI: GPT-5.4 Mini +50.4%Qwen: Qwen3.8 Flash +31.9%OpenAI: GPT-4.1 +31.3%
Top 100 by weekly usage
| # | Model | Weekly tokens | WoW change | Blended $/1M | Est. weekly spend |
|---|---|---|---|---|---|
| 1 | DeepSeek: DeepSeek V4.1 Flash | 38.8T | -32.5% | $0.131 | $5.1M |
| 2 | Z.ai: GLM 5.3 Flash | 11.6T | -9.9% | $0.237 | $2.8M |
| 3 | Xiaomi: MiMo-V2.6-Flash | 9.0T | -22.1% | $0.175 | $1.6M |
| 4 | Tencent: Hy4 preview | 8.2T | +0.3% | $1.25 | $10.2M |
| 5 | Anthropic: Claude Opus 5.5 | 6.8T | -49% | $8.00 | $54.4M |
| 6 | OpenAI: GPT-6 Luna | 6.3T | -12.5% | $0.200 | $1.3M |
| 7 | StepFun: Step 5 Preview | 5.8T | -61.3% | $1.43 | $8.3M |
| 8 | Z.ai: GLM 5.3 | 5.3T | +27.3% | $0.409 | $2.2M |
| 9 | MiniMax: MiniMax M3.1 Flash (free) | 5.0T | — | Free | $0 |
| 10 | DeepSeek: DeepSeek V4 Flash 0423 | 7.5T | -12.6% | $0.114 | $852K |
| 11 | NVIDIA: Nemotron 3 Ultra | 4.5T | -38.9% | $1.05 | $4.8M |
| 12 | Meta: Muse Spark 1.3 Contributor | 3.3T | +16.9% | $0.125 | $407K |
| 13 | Tencent: Hy3 | 3.1T | +4.5% | $0.231 | $727K |
| 14 | typesafe/jev-1.13 | 2.9T | -10.1% | — | — |
| 15 | Anthropic: Claude Sonnet 5.5 | 2.5T | -7% | $4.00 | $9.9M |
| 16 | Anthropic: Claude Haiku 5.5 | 2.0T | -34.5% | $0.200 | $396K |
| 17 | OpenAI: GPT-6.1 Sol | 1.9T | -44.6% | $4.00 | $7.6M |
| 18 | MoonshotAI: Kimi K3 | 1.8T | -91.1% | $3.16 | $5.8M |
| 19 | OpenAI: GPT-5.6 Luna | 1.7T | -9.6% | $0.450 | $765K |
| 20 | Google: Gemini 3.8 Flash | 1.7T | -17.4% | $1.65 | $2.7M |
| 21 | Xiaomi: MiMo-V2.6-Pro | 1.7T | +10.3% | $0.544 | $901K |
| 22 | MiniMax: MiniMax M3 | 1.4T | +0.6% | $0.525 | $723K |
| 23 | NVIDIA: Nemotron 3.5 Super VL | 1.3T | — | $0.237 | $309K |
| 24 | Poolside: Laguna S 2.1 | 1.0T | -26% | $0.113 | $116K |
| 25 | Upstage: Solar Mini 4 | 1.0T | -0.1% | $0.087 | $90K |
| 26 | typesafe/jev-1.14 | 1.0T | — | — | — |
| 27 | Z.ai: GLM 5.2 | 1.0T | -27.6% | $1.78 | $1.8M |
| 28 | Google: Gemini 4 Argon | 999.7B | new 🚀 | $4.00 | $4.0M |
| 29 | NVIDIA: Nemotron 3.5 Lightning | 958.1B | -21.6% | $0.102 | $98K |
| 30 | DeepSeek: DeepSeek V4 Pro 0423 | 1.6T | -21.8% | $1.19 | $1.9M |
| 31 | Qwen: Qwen3.6 35B A3B | 877.3B | +366.9% | $0.313 | $274K |
| 32 | Anthropic: Claude Opus 5 | 802.5B | -0.7% | $10.00 | $8.0M |
| 33 | Google: Gemini 3 Flash Preview | 733.3B | -63.6% | $1.13 | $825K |
| 34 | Dots Studio: Dots3-Note Preview (free) | 704.6B | -10.3% | Free | $0 |
| 35 | NVIDIA: Nemotron 3 Super | 690.0B | -5.1% | $0.825 | $569K |
| 36 | Qwen: Qwen3.8 Flash | 638.4B | +31.9% | $0.230 | $147K |
| 37 | Upstage: Solar Pro 4 | 622.8B | -35.9% | $0.263 | $163K |
| 38 | Google: Gemini 2.5 Flash Lite | 615.1B | +1.8% | $0.175 | $108K |
| 39 | Anthropic: Claude Sonnet 5 | 588.1B | -68.3% | $4.00 | $2.4M |
| 40 | Google: Gemma 4 31B | 551.8B | -6.7% | $0.250 | $138K |
| 41 | OpenAI: gpt-oss-120b | 544.3B | -39.2% | $0.263 | $143K |
| 42 | Xiaomi: MiMo-V2.5 | 509.0B | -4.7% | $0.149 | $76K |
| 43 | Google: Gemini 3.1 Flash Lite | 479.5B | -95.1% | $0.563 | $270K |
| 44 | OpenAI: GPT-6 Sol | 455.2B | -35.9% | $4.00 | $1.8M |
| 45 | OpenAI: GPT-5.6 Sol | 451.3B | -52.5% | $8.00 | $3.6M |
| 46 | Qwen: Qwen3.8 27B | 426.4B | +1.6% | $0.367 | $157K |
| 47 | OpenAI: GPT-6 Luna Pro | 387.5B | +12% | $0.200 | $77K |
| 48 | Google: Gemini 3.7 Flash | 383.6B | -50.3% | $1.50 | $575K |
| 49 | OpenAI: GPT-6 Astra | 374.6B | -41.5% | $20.00 | $7.5M |
| 50 | Google: Gemma 4 26B A4B | 370.7B | -15.8% | $0.142 | $53K |
| 51 | Qwen: Qwen3.7 Flash | 339.1B | -8.8% | $0.055 | $19K |
| 52 | Google: Gemini 2.5 Flash | 332.8B | -25.3% | $0.850 | $283K |
| 53 | Anthropic: Claude Fable 5.1 | 290.6B | -43% | $20.00 | $5.8M |
| 54 | Thinking Machines: Inkling | 288.8B | -41.1% | $1.76 | $509K |
| 55 | DeepSeek: DeepSeek V3.2 | 287.3B | -10.6% | $0.315 | $90K |
| 56 | xAI: Grok 4.7 | 285.4B | -55.7% | $3.00 | $856K |
| 57 | anthropic/claude-4.6-sonnet | 272.7B | -9.1% | — | — |
| 58 | Mistral: Mistral Nemo | 268.3B | +9.3% | $0.029 | $8K |
| 59 | Google: Gemini 3.5 Flash Lite | 225.1B | -19.2% | $0.850 | $191K |
| 60 | OpenAI: GPT-5.6 Terra | 222.7B | -2.6% | $4.50 | $1.0M |
| 61 | Meta: Muse Wonder | 219.0B | +203.8% | $8.00 | $1.8M |
| 62 | OpenAI: GPT-4o-mini | 218.8B | +4.5% | $0.263 | $57K |
| 63 | Qwen: Qwen3.8 Max Prime | 207.5B | -29.8% | $6.00 | $1.2M |
| 64 | Cohere: North Mini Code (free) | 198.9B | +13.3% | Free | $0 |
| 65 | OpenAI: gpt-oss-20b | 196.7B | -46.1% | $0.036 | $7K |
| 66 | qwen/qwen3-embedding-8b | 195.6B | +11.1% | — | — |
| 67 | anthropic/claude-4.5-sonnet | 184.4B | +23.1% | — | — |
| 68 | anthropic/claude-4.8-opus | 178.0B | -21.2% | — | — |
| 69 | OpenAI: GPT-5.4 Mini | 173.0B | +50.4% | $1.69 | $292K |
| 70 | Xiaomi: MiMo-V2.5-Pro | 161.7B | -7.4% | $0.381 | $62K |
| 71 | Mistral: Mistral Large 4 | 161.0B | -16.6% | $2.06 | $332K |
| 72 | anthropic/claude-4.5-haiku | 156.1B | winding down | — | — |
| 73 | Thinking Machines: Inkling Small | 150.5B | -0.5% | $0.637 | $96K |
| 74 | openai/text-embedding-3-small | 146.8B | +0.8% | — | — |
| 75 | xAI: Grok 4.6 | 137.1B | +17.8% | $3.00 | $411K |
| 76 | Google: Gemini 3.1 Pro Preview | 135.9B | -64.7% | $4.50 | $612K |
| 77 | inclusionAI: Ling 3.0 Flash | 135.9B | +6.1% | $0.032 | $4K |
| 78 | Z.ai: GLM 5.3 FlashX | 132.9B | -44.6% | $0.590 | $78K |
| 79 | MoonshotAI: Kimi K2.6 | 132.2B | -5.9% | $1.34 | $177K |
| 80 | OpenAI: GPT-5 | 125.6B | -13.4% | $3.44 | $432K |
| 81 | inclusionAI: Ling 3.1 Flash | 119.0B | -82.1% | $0.113 | $13K |
| 82 | voyageai/rerank-2.5-lite | 115.4B | +6.8% | — | — |
| 83 | Xiaomi: MiMo-V2.6-Pro-UltraSpeed | 114.7B | -18.2% | $5.44 | $624K |
| 84 | Google: Gemini 3.1 Flash Lite Preview | 105.1B | +1.1% | $0.563 | $59K |
| 85 | Meta: Muse Spark Open | 100.2B | — | $2.00 | $200K |
| 86 | OpenAI: GPT-5.4 Nano | 95.4B | -10.6% | $0.463 | $44K |
| 87 | OpenAI: GPT-5.6 Luna Pro | 94.2B | -10% | $0.450 | $42K |
| 88 | anthropic/claude-4.6-opus | 89.6B | -15.8% | — | — |
| 89 | Meta: Muse Spark 1.3 | 88.2B | -84.1% | $2.00 | $176K |
| 90 | OpenAI: GPT-4.1 | 85.1B | -16.4% | $3.50 | $298K |
| 91 | Reflection: Beam 1.0 501B A23B | 84.8B | — | $0.475 | $40K |
| 92 | Poolside: Laguna XS 2.1 | 82.6B | -42.9% | $0.075 | $6K |
| 93 | OpenAI: GPT-4.1 | 81.7B | +31.3% | $3.50 | $286K |
| 94 | DeepSeek: DeepSeek V4 Flash Vision Exp | 79.5B | -14.8% | $0.323 | $26K |
| 95 | Google: Gemini 3.6 Flash | 79.1B | -7.4% | $1.50 | $119K |
| 96 | Qwen: Qwen3.5-Flash | 77.3B | +18.1% | $0.114 | $9K |
| 97 | perplexity/pplx-decider-v1.1-27b | 73.0B | -60.4% | — | — |
Spend estimates assume the blended 3:1 input:output ratio applies to each model's token mix — real workloads vary. 11 ranked models aren't in our pricing dataset (free variants, deprecated slugs) and show "—". Usage data © OpenRouter, CC BY 4.0.