OpenRouter
OpenRouter is one API and one bill for more than 600 AI models from 80+ providers, including GPT, Claude, Gemini, DeepSeek and Qwen, at the providers' own token prices plus a 5.5% fee on credit purchases, as of Nov 6, 2026. Its public rankings show which models get the most use.
OpenRouter is a service that gives developers one API key, one bill and one OpenAI-style API for 600+ models from 80+ providers, open and closed: OpenAI, Anthropic, Google, DeepSeek, Qwen, Mistral and many more. It charges each model at the provider's own price with no markup on inference, and takes a 5.5% fee when you buy credits (as of Nov 6, 2026). A set of free models allows 50 requests a day, or 1,000 after a $10 purchase. Its public rankings of tokens processed are a widely watched measure of which models developers actually use.
Pick OpenRouter to switch models without new accounts, to compare models on the same prompt, or to fall back to another provider when one is down. If you use just one maker's models, its own API (OpenAI, Anthropic, Google) avoids the credit fee and may get new features first. For open models only, Hugging Face Inference Providers works the same way with no fee on credits, and for private use with no API at all, run models on your own computer with Ollama or LM Studio.
Quick facts
| Maker | OpenRouter |
| Get it | openrouter.ai: web chat, API, SDKs |
| Price | Provider prices plus a 5.5% fee on credit purchases (8% on the Business plan); free models (as of Nov 6, 2026) |
| Models | 600+ from 80+ providers (Nov 6, 2026) |
| Free tier | 25+ free models, 50 requests a day; 1,000 a day after buying $10 of credits |
| API | OpenAI compatible |
| Best for | One key for many models, comparing and switching models |
How OpenRouter pricing works
| Plan | Fee | What you get |
|---|---|---|
| Free | None | 25+ free models from 4 providers, 50 requests a day, 20 a minute |
| Pay as you go | 5.5% on credit purchases | All 600+ models, provider rate limits, 1,000 free-model requests a day after $10 |
| Business | 8% on credit purchases | Pay as you go for teams, up to 1,000 workspaces |
| Enterprise | Contract, fee discounts | Invoicing, higher limits |
Source: OpenRouter pricing, read Nov 6, 2026. Token prices themselves match each provider's. Bring your own key (BYOK): you can use your own provider accounts through OpenRouter, free up to $25,000 of usage a month, then at 5% of what the same calls would cost on OpenRouter.
Features
One API for every model. Code written for OpenAI's API works by changing the address and the model name, for example to a Claude, Gemini or DeepSeek model. That makes it easy to test one prompt on many models or to route cheap work to a cheap model.
Provider routing and fallbacks. Popular open models are served by several providers at different prices and speeds; OpenRouter picks among them and moves to another when one fails, with limits set by each provider.
Rankings. OpenRouter publishes which models, authors and apps use the most tokens through it, updated daily. These measure what its customers use, not quality, and they lean toward cheap models used in high volume by coding agents. See AI Leaderboards for quality rankings.
Chat. A web chat lets you try models and compare them side by side before using them in code.
OpenRouter rankings (Nov 2026)
| Rank | Model | Maker | Tokens |
|---|---|---|---|
| 1 | DeepSeek V4.1 Flash 1017 | DeepSeek | 56.5 trillion |
| 2 | DeepSeek V4.1 Flash | DeepSeek | 16.5 trillion |
| 3 | GLM 5.3 Flash | Z.ai | 15 trillion |
| 4 | GPT-6 Luna | OpenAI | 7.26 trillion |
| 5 | Step 5 Preview | StepFun | 6.67 trillion |
| 6 | Hy4 preview | Tencent | 6.58 trillion |
| 7 | Claude Opus 5.5 | Anthropic | 6.53 trillion |
Source: OpenRouter rankings, usage data through Nov 5, 2026, read Nov 6, 2026. By share of text requests in the week from Oct 26, 2026, DeepSeek led with 29.8%, then OpenAI 18.8%, Google 14.4%, Z.ai (GLM) 11.9%, Anthropic 5.7% and Qwen 5.6%. The busiest apps that day were the Hermes Agent and Claude Code, followed by other coding agents including Kilo Code and Codex.
OpenRouter vs the alternatives
- Hugging Face Inference Providers: one OpenAI-compatible API for open models only, at provider rates with no fee on credits; PRO includes $2 of credits a month.
- The makers' own APIs: OpenAI, Anthropic and Google sell direct with no credit fee, plus features such as batch discounts that a router may not pass on.
- Ollama: local models free, and cloud open models on its own subscription plans.
- Poe: a consumer app with many models under one subscription, for chatting rather than building.
Frequently asked questions
Is OpenRouter free? OpenRouter offers 25+ free models with a limit of 50 requests a day (1,000 after buying $10 of credits); paid models cost the provider's price plus a 5.5% fee on credit purchases, as of Nov 6, 2026.
Does OpenRouter charge more than the providers? OpenRouter passes through each provider's token prices with no markup on inference, but adds a 5.5% fee when you buy credits (8% on Business).
How many models does OpenRouter have? OpenRouter lists more than 600 models from more than 80 providers as of Nov 6, 2026.
What is the most used model on OpenRouter? DeepSeek V4.1 Flash 1017 led OpenRouter's rankings by tokens processed, with 56.5 trillion, in data through Nov 5, 2026.
Is OpenRouter safe for private data? OpenRouter sends your prompts to the provider that serves each model, so each provider's data policy applies too; for data that must stay with you, run open models locally with Ollama or LM Studio.
Sources
- Pricing, OpenRouter, read Nov 6, 2026
- FAQ, OpenRouter docs, read Nov 6, 2026
- Rankings, OpenRouter, read Nov 6, 2026
- OpenRouter, read Nov 6, 2026
- What is OpenRouter | All about OpenRouter in 10 minutes, codebasics, Mar 25, 2026 (video)
Moments
What is OpenRouter | All about OpenRouter in 10 minutes
codebasics
- OpenRouter is a service that gives developers one API key, one bill and one OpenAI-style API for 600+ models from 80+…
- Pick OpenRouter to switch models without new accounts, to compare models on the same prompt, or to fall back to…
- One API for every model. Code written for OpenAI's API works by changing the address and the model name, for example…
- Rankings. OpenRouter publishes which models, authors and apps use the most tokens through it, updated daily. These…
- Chat. A web chat lets you try models and compare them side by side before using them in code.
Photos
See also
-
Hugging FaceHugging Face is the main hub for open AI, hosting more than 2 million models, 500,000 datasets and 1 million Spaces apps; it is free to use, with PRO at $9 a month. It also runs Inference Providers, one API for hosted open models, and the HuggingChat assistant. -
OllamaOllama is a free, open-source tool for downloading and running open AI models such as Gemma, Qwen, gpt-oss and DeepSeek on your own Mac, Windows or Linux computer. Its optional cloud models are pay per token, with Pro at $20 a month and Max at $100, as of Nov 6, 2026. -
LM StudioLM Studio is a free desktop app for Mac, Windows and Linux that finds, downloads and runs open AI models such as Gemma, Qwen and gpt-oss on your own computer, free for work use since Jul 2025. Its newer agent app, Bionic, adds paid US-hosted models from $20 a month. -
AI LeaderboardsAI leaderboards rank models by human votes (LMArena), by a mix of tests (Artificial Analysis), or by single tasks such as coding (SWE-bench, Terminal-Bench) and hard exam questions (Humanity's Last Exam). On Nov 6, 2026, Google's Gemini 4 Carbon led both LMArena Text and the Artificial Analysis Intelligence Index, and Claude Opus 5.5 led coding.
-
DeepSeekDeepSeek is a Chinese AI lab whose free chat app and open-weight models, now the V4 family (V4-Pro and V4.1-Flash), are MIT licensed and among the cheapest to use by API. Its privacy policy says user data is stored in China, which led Italy's privacy regulator to block it there in Jan 2025. -
GLMGLM is the model family from Zhipu AI, which trades as Z.ai, with a free chatbot at chat.z.ai. The newest, GLM-5.3 and GLM-5.3-Flash (Aug 25, 2026), are open weight, and the GLM Coding Plan for coding agents starts at $18 a month, as of Nov 6, 2026.