Ranked by real token throughput on the OpenRouter marketplace - not benchmarks. Useful for seeing which models people actually run.
Tokens
120.05T
Requests
5.6B
Models
463
Lead
Hy4 preview
DeepSeek
Token mix this window
97% prompt · 3% completion
Share of token volume this window
14.8%
↑31%
12.1%
↑29%
10.1%
↓2%
10.0%
↑1%
5.1%
↑92%
3.9%
↓12%
3.0%
↓8%
2.9%
↓27%
2.3%
↑6%
2.1%
↑295%
Showing 20 of 463
| # | Model | Type | Tokens | Share | Requests | Tools | Price / M | Context | Δ wk |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Hy4 preview Tencent · OpenRouter | Language | 17.8T 17.6T in · 159.1B out | 14.8% | 151.9M | - | $0.83 / $2.50 | 1.0M | ↑31% |
| 2 | GPT-5.6 Luna OpenAI · OpenRouter | Vision | 14.6T 14.1T in · 495.5B out | 12.1% | 725.4M | - | $0.20 / $1.20 | 1.1M | ↑29% |
| 3 | GLM 5.3 Flash Z.ai · OpenRouter | Vision | 12.1T 11.7T in · 371.7B out | 10.1% | 305.2M | - | $0.07 / $0.25 | 1.3M | ↓2% |
| 4 | DeepSeek V4 Flash 0731 DeepSeek · OpenRouter | Language | 12.0T 11.6T in · 386.9B out | 10.0% | 533.4M | - | $0.04 / $0.08 | 1.3M | ↑1% |
| 5 | MiMo-V2.5 Xiaomi · OpenRouter | Audio | 6.08T 6.03T in · 48.3B out | 5.1% | 80.9M | - | $0.14 / $0.28 | 1.1M | ↑92% |
| 6 | DeepSeek V4 Flash 0423 DeepSeek · OpenRouter | Language | 4.63T 4.38T in · 246.9B out | 3.9% | 497.3M | - | $0.07 / $0.13 | 1.0M | ↓12% |
| 7 | Nemotron 3 Ultra (free) NVIDIA · free · OpenRouter | Language | 3.57T 3.55T in · 18.0B out | 3.0% | 28.5M | - | Free | 1M | ↓8% |
| 8 | Hy3 Tencent · OpenRouter | Language | 3.51T 3.45T in · 57.7B out | 2.9% | 62.2M | - | $0.13 / $0.53 | 262K | ↓27% |
| 9 | GLM 5.3 Z.ai · OpenRouter | Language | 2.79T 2.72T in · 70.2B out | 2.3% | 39.1M | - | $1.40 / $4.40 | 1.3M | ↑6% |
| 10 | Gemini 3.8 Flash Google · OpenRouter | Audio | 2.55T 2.51T in · 44.1B out | 2.1% | 49.4M | - | $0.75 / $3.75 | 1.0M | ↑295% |
| 11 | DeepSeek V4.1 Flash DeepSeek · OpenRouter | Vision | 2.22T 2.18T in · 31.6B out | 1.8% | 28.7M | - | $0.15 / $0.60 | 1.0M | - |
| 12 | GLM 5.2 Z.ai · OpenRouter | Language | 2.03T 1.99T in · 46.5B out | 1.7% | 53.5M | - | $0.60 / $2.00 | 1.0M | ↓10% |
| 13 | Audio | 1.97T 1.94T in · 32.3B out | 1.6% | 24.7M | - | $0.10 / $0.20 | 1.0M | ↑494% | |
| 14 | GPT-5.6 Sol OpenAI · OpenRouter | Vision | 1.70T 1.68T in · 21.3B out | 1.4% | 33.4M | - | $2.00 / $10 | 1.1M | ↓7% |
| 15 | Solar Pro 4 Upstage · OpenRouter | Language | 1.68T 1.66T in · 16.0B out | 1.4% | 26.5M | - | $0.09 / $0.36 | 524K | ↑70% |
| 16 | Kimi K3 Moonshot · OpenRouter | Vision | 1.61T 1.59T in · 23.5B out | 1.3% | 26.9M | - | $2.30 / $12 | 1.0M | ↓15% |
| 17 | Claude Opus 5 Anthropic · OpenRouter | Vision | 1.55T 1.52T in · 29.6B out | 1.3% | 21.2M | - | $5.00 / $25 | 1M | - |
| 18 | MiniMax M3 MiniMax · OpenRouter | Vision | 1.47T 1.45T in · 22.8B out | 1.2% | 28.2M | - | $0.30 / $1.20 | 1.0M | ↑1% |
| 19 | Claude Sonnet 5 Anthropic · OpenRouter | Vision | 1.39T 1.36T in · 22.3B out | 1.2% | 27.1M | - | $2.00 / $10 | 1M | ↑7% |
| 20 | DeepSeek V4 Pro 0423 DeepSeek · OpenRouter | Language | 1.19T 1.12T in · 63.6B out | 1.0% | 46.5M | - | $0.80 / $1.60 | 1.0M | ↓16% |
Weekly request share by model author
Models that actually invoke tools
Images processed through OpenRouter
Public apps by token volume this week
Hermes Agent
Hermes Agent is an open-source, self-improving AI agent by Nous Research that runs persistently with memory across sessions, and builds reusable skills from experience. It comes with 40+ built-in tools, including web search, browser automation, and vision, plus scheduled automations and subagents.
Claude Code
Claude Code is Anthropic's agentic coding tool that reads your entire codebase, plans and executes changes across files, runs tests, and iterates on failures, all from natural language prompts.
Cline
Cline is an open-source AI coding agent that lives inside your IDE, autonomously exploring your codebase, editing files, running terminal commands, and using browser automation.
Kilo Code
Kilo Code is an open-source AI coding agent that works across VS Code, JetBrains, and CLI to help developers ship code faster with agentic workflows.
pi
There are many coding agents, but this one is yours.
omp
OpenClaw
OpenClaw is an open-source AI agent that connects to your messaging apps and takes real actions on your behalf, from running commands and browsing the web to managing files and sending emails.
Codex
A coding agent that helps you build and ship with AI
Freebuff
Powerful coding models, funded by ads
DeepSeek Harness
Open WebUI
Open WebUI is a self-hosted AI platform providing a chat interface for Large Language Models.
Framer
Create a professional website with Framer’s no-code AI website builder. Design freely, manage CMS content, optimize SEO, collaborate, and publish fast.
Source: OpenRouter rankings, as of 2026-09-11. Licensed under CC BY 4.0. Price is prompt / completion per million tokens. First-party app and direct vendor traffic are not included.