Free tier - verified
Is the OpenRouter API free?
Yes. The OpenRouter API offers free model variants whose IDs end in :free. Whether a card is required is not stated. Whether your data is used for training depends on the provider serving the model.
Read from openrouter.ai. Compare every free LLM API
- Offer
- Free tier
- Card required
- Not stated
- Trains on data
- Depends on the serving provider
- Verified
- 2026-10-18
The offer
- Offer class
Standing free tier
- How to get it
Create an OpenRouter account and API key, then call a model ID ending in ':free' (or the 'openrouter/free' Free Models Router). No purchase is needed for the 50 requests/day level.
“Append :free to the ID of a model that has a free entry.”
openrouter.ai “All new users receive a small free allowance to test out OpenRouter. There are many free models available on OpenRouter. These models have low rate limits (50 requests per day total) and are usually not suitable for production use. If you have purchased at least 10 credits, the free models will be limited to 1000 requests per day.”
openrouter.ai - Card required
Not stated
Not stated. The free-variant limits table includes a 'Less than 10' credits-purchased row, i.e. use without any purchase; the 1000 RPD level requires buying at least 10 credits. The 'small free allowance' for new users is not quantified (treat as unspecified credits).
Free-tier limits
Values are copied exactly as the provider publishes them, units included. Not stated means the public docs give no value.
| Model | Requests per minute | Requests per day | Context | Served byOpenRouter's label |
|---|---|---|---|---|
| nvidia/nemotron-3-ultra-550b-a55b:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 1,000,000 | Nvidia: 'Prompts are retained for unknown period | May train' |
| nvidia/nemotron-3.5-super-vl:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 1,000,000 | Nvidia: 'Prompts are retained for unknown period | May train' |
| nvidia/nemotron-3-super-120b-a12b:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 262,144 | Nvidia: 'Prompts are retained for unknown period | May train' |
| nvidia/nemotron-3.5-lightning:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 1,000,000 | Nvidia: 'Prompts are retained for unknown period | May train' |
| thinkingmachines/inkling:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 1,048,576 | Thinking Machines: 'Prompts are retained for unknown period | May train' |
| thinkingmachines/inkling-small:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 1,048,576 | Thinking Machines: 'Prompts are retained for unknown period | May train' |
| google/gemma-4-31b-it:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 262,144 | Google AI Studio: 'Retained for 55 days | Does not train' |
| google/gemma-4-26b-a4b-it:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 262,144 | Google AI Studio: 'Retained for 55 days | Does not train' |
| poolside/laguna-s-2.1:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 262,144 | Poolside: 'Prompts are retained for unknown period | Does not train' |
| poolside/laguna-xs-2.1:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 262,144 | Poolside: 'Prompts are retained for unknown period | Does not train' |
| cohere/north-mini-code:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 256,000 | Cohere: 'Retained for 30 days | Does not train' |
| dots-studio/dots-3-note-preview:free | 20 | 50 (less than 10 credits purchased all time) / 1000 (at least 10 credits purchased) | 512,000 | AtlasCloud: 'Prompts are retained for unknown period | Does not train' |
Platform-wide free-variant limits; OpenRouter publishes no per-model free limits.
16 model IDs end in ':free' in the models API today (471 models total), plus 'openrouter/free' (Free Models Router, $0). Omitted here: xingchen-agi/xing4.0-29b-a4b:free, liquid/lfm-2.5-2.6b:free (served by Liquid, 'May train' per table), nvidia/nemotron-3.5-content-safety:free, nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free. Each listed free model has a single endpoint/provider today (from /api/v1/models/{id}/endpoints). The free list changes often.
From the limits docs
“| Credits purchased (all time) | Requests per minute | Requests per day | | Less than 10 | 20 | 50 | | At least 10 | 20 | 1000 |”
“To absorb rounding and top-up fees, the higher daily ceiling is granted starting one credit below the table's threshold (currently 9 credits)”
Does it train on your data?
Depends on the serving provider
Set by the provider serving each free model
OpenRouter itself does not store or use prompts unless you opt in. Training depends on the provider serving the free endpoint: per OpenRouter's provider table, Nvidia and Thinking Machines (which serve 6 of the 12 listed free models) 'May train'; Google AI Studio, Poolside, Cohere and AtlasCloud 'Does not train'. Account privacy settings have separate toggles for free and paid models; opting out of training stops routing to providers that train.
“On your account settings page, you can set whether you would like to allow routing to providers that may train on your data (according to their own policies). There are separate settings for paid and free models.Wherever possible, OpenRouter works with providers to ensure that prompts will not be trained on, but there are exceptions. If you opt out of training in your account settings, OpenRouter will not route to providers that train. This setting has no bearing on OpenRouter's own policies and what we do with your prompts.”
“OpenRouter does not store your prompts or responses, unless you opt in to one or both of the following:”
“OpenRouter Use of Inputs/Outputs: Allow OpenRouter to use your prompt and completion data to improve the product. In exchange, you receive a 1% discount on all model usage. Off by default.”
“Providers that do log, or where we have been unable to confirm their policy, will not be routed to unless the model training toggle is switched on in the privacy settings tab.”
Caveat: The provider table does not distinguish free endpoints from paid endpoints of the same provider; the 'Google AI Studio | Does not train' row is OpenRouter's characterisation and differs from Google's own unpaid-quota terms (OpenRouter's account with Google is not described).
Also read: https://openrouter.ai/docs/guides/privacy/data-collection, https://openrouter.ai/docs/faq
Restrictions
Not for production
“These models have low rate limits (50 requests per day total) and are usually not suitable for production use.”
openrouter.ai Negative balance
“If your account has a negative credit balance, you may see 402 errors, including for free models.”
openrouter.ai Extra accounts
“Making additional accounts or API keys will not affect your rate limits, as we govern capacity globally.”
openrouter.ai Availability
“Free variants provide access to models without cost, but may have different rate limits or availability compared to paid versions.”
openrouter.ai
Upgrade path
Buying at least 10 credits (5.5% fee, $0.80 minimum) raises the free-variant cap from 50 to 1000 requests per day; the paid variant of a model has no platform-level request cap.
“On free variants, purchase at least 10 credits to raise your daily limit, or switch to the paid variant of the model, which has no platform-level request cap.”
FAQ
Is the OpenRouter API free?
Yes. The OpenRouter API offers free model variants whose IDs end in :free. Whether a card is required is not stated. Whether your data is used for training depends on the provider serving the model.
What are the OpenRouter free limits?
20 requests per minute; 50 requests per day, or 1000 after buying at least 10 credits. 16 model IDs end in ':free' in the models API today (471 models total), plus 'openrouter/free' (Free Models Router, $0). Omitted here: xingchen-agi/xing4.0-29b-a4b:free, liquid/lfm-2.5-2.6b:free (served by Liquid, 'May train' per table), nvidia/nemotron-3.5-content-safety:free, nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free. Each listed free model has a single endpoint/provider today (from /api/v1/models/{id}/endpoints). The free list changes often.
Does OpenRouter train on your data?
Depends on the serving provider (Set by the provider serving each free model). The terms say: "On your account settings page, you can set whether you would like to allow routing to providers that may train on your data (according to their own policies). There are separate settings for paid and free models.Wherever possible, OpenRouter works with providers to ensure that prompts will not be trained on, but there are exceptions. If you opt out of training in your account settings, OpenRouter will not route to providers that train. This setting has no bearing on OpenRouter's own policies and what we do with your prompts."
Sources
Last verified from OpenRouter's own pages below. Quotes are verbatim.
Newsletter
Know when a free tier changes
One email per week. New model releases, pricing and limit changes, and what they mean for what you build.