OpenRouter free models: the current list and limits
Updated · 2 min read
Any model on OpenRouter whose ID ends in :free costs nothing per token. One OpenRouter key works in every app that supports OpenRouter, so this is one of the easiest ways to try BYOK apps for free. The list changes often, so the table below is pulled from OpenRouter's public models API and refreshed every day.
16 free models on OpenRouter, from its public models API (updated daily, last on ), newest first:
| Model | Model ID to paste in your app | Context | Input |
|---|---|---|---|
| Free Models Router | openrouter/free | 200K | Text, Image |
| Recursive: Flint | recursive/flint-v1.0:free | 262K | Text |
| XingChen: Xing4.0 29B A4B | xingchen-agi/xing4.0-29b-a4b:free | 262K | Text |
| NVIDIA: Nemotron 3.5 Super VL | nvidia/nemotron-3.5-super-vl:free | 1M | Text, Image, Video |
| Dots Studio: Dots3-Note Preview | dots-studio/dots-3-note-preview:free | 512K | Text, Image |
| LiquidAI: LFM2.5-2.6B | liquid/lfm-2.5-2.6b:free | 66K | Text |
| NVIDIA: Nemotron 3.5 Lightning | nvidia/nemotron-3.5-lightning:free | 1M | Text |
| Thinking Machines: Inkling Small | thinkingmachines/inkling-small:free | 1M | Text, Image, Audio |
| Poolside: Laguna S 2.1 | poolside/laguna-s-2.1:free | 262K | Text |
| Thinking Machines: Inkling | thinkingmachines/inkling:free | 1M | Text, Image, Audio |
| Poolside: Laguna XS 2.1 | poolside/laguna-xs-2.1:free | 262K | Text |
| Cohere: North Mini Code | cohere/north-mini-code:free | 256K | Text |
| NVIDIA: Nemotron 3.5 Content Safety | nvidia/nemotron-3.5-content-safety:free | 128K | Text, Image |
| NVIDIA: Nemotron 3 Ultra | nvidia/nemotron-3-ultra-550b-a55b:free | 1M | Text |
| NVIDIA: Nemotron 3 Nano Omni | nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free | 256K | Text, Audio, Image, Video |
| Google: Gemma 4 26B A4B | google/gemma-4-26b-a4b-it:free | 262K | Image, Text, Video |
| Google: Gemma 4 31B | google/gemma-4-31b-it:free | 262K | Image, Text, Video |
openrouter/free is not a single model but a router: it picks a free model at random for each request, only among models that support what the request needs (such as images, tool calling or structured output). It is a good default when a particular free model is busy.
Free limits
From OpenRouter's limits documentation on 2026-10-29:
| Credits you have ever bought | Requests per minute | Requests per day |
|---|---|---|
| Less than $10 | 20 | 50 |
| $10 or more | 20 | 1,000 |
The daily count resets at midnight UTC. Buying $10 once raises the daily limit for good, and those credits stay in your balance for paid models (unused credits can expire after a year). Two more things to know:
- A negative balance blocks free models too. If paid usage pushed your balance below zero, top up before free models work again.
- Check what's left by calling
https://openrouter.ai/api/v1/keywith your key; thefree_model_daily_requestsfield shows today's count and limit.
Use free models in an app
- Create an OpenRouter key (no payment needed). Steps: how to get an OpenRouter API key.
- Paste the key into the app's OpenRouter settings. Apps without an OpenRouter option often accept a custom OpenAI-compatible provider: base URL
https://openrouter.ai/api/v1. - Pick a model ID ending in
:freefrom the table above, oropenrouter/free. Many apps load OpenRouter's model list for you; if yours asks you to type the ID, copy it exactly, including:free.
Privacy
OpenRouter passes your prompts to the provider that hosts the model, and some providers may train on what you send. In your OpenRouter privacy settings there are separate switches for paid and free models that control whether OpenRouter may route to providers that train on data. If you turn that off, OpenRouter skips those providers, which can leave fewer free models available to you. Either way, don't send private or confidential material to free endpoints.
The catch
- The list changes. Free models come and go, often previews that later become paid. If a saved model ID stops working, check the table above for a replacement.
- Free endpoints can be slow or busy at peak times, and 50 requests a day runs out quickly in a chat app that sends several requests per message (titles, summaries, retries).
- No frontier commercial models. You get capable open models, not GPT-6, Claude or Gemini Pro.
Apps where you can use OpenRouter's free models
VisitSideNote Pro
30-day free trial, then $49 one-time · Windows
Docked AI assistant for Windows 10 and 11 that works with your own OpenAI, Anthropic, Gemini, DeepSeek or OpenRouter key, or local Ollama and LM Studio models.
Works with:OpenAIOpenRouterAnthropicGoogle Gemini+3 more
VisitOpenDocBot
Free for up to 30 users; commercial license above that
Client-side AI agent for Word, Excel and PowerPoint that runs on your own OpenAI, Anthropic, Gemini, DeepSeek or OpenRouter key, or a local Ollama model.
Works with:OpenAIAnthropicGoogle GeminiDeepSeek+2 more


VisitAnythingLLM
Free · macOS, Windows, Linux · Open source
Private AI app for chatting with your documents. Runs local models or connects to OpenAI, Anthropic, Gemini and others with your key.
Works with:OpenAIAzure OpenAIAnthropicGoogle Gemini+9 more

Other free options
Google's Gemini free tier offers stronger commercial models for free, Groq's free plan serves gpt-oss very fast, and local models are free and private on your own computer. All options side by side: free AI API keys.
Questions
Are OpenRouter's free models really free?
Yes. Models whose ID ends in :free cost nothing per token. They have request limits instead: 20 per minute, and 50 per day until you have bought $10 of credits in total, then 1,000 per day.
Do I need to add credits to use free models?
No, a new account can use them right away. Buying $10 of credits once raises the daily limit from 50 to 1,000 requests, and the credits stay available for paid models. If your balance goes negative, even free models return errors until you top up.
Is there a free GPT, Claude or Gemini model on OpenRouter?
Not on 2026-10-29. The free list holds open models, such as Google's Gemma 4 and NVIDIA's Nemotron, not OpenAI's, Anthropic's or Google's commercial models. For a free Gemini, use Google's own free tier.
Why does a free model say it is busy or rate limited?
Free endpoints are shared and can be overloaded at busy times, and you may have hit your own per-minute or daily limit. Wait a minute, try another free model, or use the openrouter/free router, which picks an available free model for you.
How do I see how many free requests I have left?
Call https://openrouter.ai/api/v1/key with your key. The free_model_daily_requests field shows used, limit and remaining for the current UTC day.
Sources
Checked against the providers' own documentation on 2026-10-29. Prices and limits change; the links below have the current figures.