AI Perks curates and provides access to exclusive discounts, credits, and deals on AI tools, cloud services, and APIs to help startups and developers save money.
What free models does OpenRouter list right now?
The models API listed 467 models on 2026-10-25, and 20 of them were free (zero prompt and completion price). The full list with context lengths is below, and it rotates as providers change what they offer.
Context lengths in this list run from 65,536 tokens up to 1,048,576 tokens. Six of the 20 are NVIDIA models and four are Google models. OpenRouter says free availability changes frequently, so check the OpenRouter models page before you build on any single model.
| Model | Model ID | Context (tokens) |
|---|---|---|
| Cohere: North Mini Code (free) | cohere/north-mini-code:free | 256,000 |
| Dots Studio: Dots3-Note Preview (free) | dots-studio/dots-3-note-preview:free | 512,000 |
| Google: Gemma 4 26B A4B (free) | google/gemma-4-26b-a4b-it:free | 262,144 |
| Google: Gemma 4 31B (free) | google/gemma-4-31b-it:free | 262,144 |
| Google: Lyria 3 Clip Preview | google/lyria-3-clip-preview | 1,048,576 |
| Google: Lyria 3 Pro Preview | google/lyria-3-pro-preview | 1,048,576 |
| LiquidAI: LFM2.5-2.6B (free) | liquid/lfm-2.5-2.6b:free | 65,536 |
| NVIDIA: Nemotron 3 Nano Omni (free) | nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free | 256,000 |
| NVIDIA: Nemotron 3 Super (free) | nvidia/nemotron-3-super-120b-a12b:free | 262,144 |
| NVIDIA: Nemotron 3 Ultra (free) | nvidia/nemotron-3-ultra-550b-a55b:free | 1,000,000 |
| NVIDIA: Nemotron 3.5 Content Safety (free) | nvidia/nemotron-3.5-content-safety:free | 128,000 |
| NVIDIA: Nemotron 3.5 Lightning (free) | nvidia/nemotron-3.5-lightning:free | 1,000,000 |
| NVIDIA: Nemotron 3.5 Super VL (free) | nvidia/nemotron-3.5-super-vl:free | 1,000,000 |
| Free Models Router | openrouter/free | 200,000 |
| Poolside: Laguna S 2.1 (free) | poolside/laguna-s-2.1:free | 262,144 |
| Poolside: Laguna XS 2.1 (free) | poolside/laguna-xs-2.1:free | 262,144 |
| Recursive: Flint (free) | recursive/flint-v1.0:free | 262,144 |
| Thinking Machines: Inkling Small (free) | thinkingmachines/inkling-small:free | 1,048,576 |
| Thinking Machines: Inkling (free) | thinkingmachines/inkling:free | 1,048,576 |
| XingChen: Xing4.0 29B A4B (free) | xingchen-agi/xing4.0-29b-a4b:free | 262,144 |
Seven models list a context of 1,000,000 tokens or more. The two Lyria entries are audio models (text and image in, text and audio out) with no :free suffix in their IDs, but the API lists them at zero price.
Top AI Credits for Startups
Apply directly through these verified programs.
OpenRouter Up to $3,000 + 0% fees for 12 months
Eligible for early-stage startups
Claude (via program) $100,000
Eligible for early-stage startups
OpenAI (via program) $2,500
Eligible for early-stage startups
Anthropic $20,000
Eligible for early-stage startups
AWS $300,000
Eligible for early-stage startups
Google Cloud $350,000
Eligible for early-stage startups
What are the OpenRouter free model limits?
Free models are capped at 20 requests per minute and 50 requests per day. The daily cap rises to 1,000 once your account has bought at least $10 of credits.
That is a 20x jump in daily requests (1,000 divided by 50) for a $10 purchase. The per-minute cap stays at 20 in both tiers.
| Account state | Per minute | Per day |
|---|---|---|
| Free models, under $10 of credits bought | 20 | 50 |
| Free models, at least $10 of credits bought | 20 | 1,000 |
| Paid variant of a model | No plan-based limit | No platform-level request cap |
Two details catch people out. Failed attempts count toward the daily cap, so a retry loop on a broken request burns your 50 fast. A negative balance also blocks free models until you add credits.
Creating extra accounts or API keys does not help. OpenRouter states that additional accounts or keys will not affect your rate limits, because capacity is governed globally. For a deeper breakdown of the tiers, see the OpenRouter free tier guide.
How does the openrouter/free router pick a model?
The Free Models Router (openrouter/free) selects a free model at random from the free models currently available. You cannot control which one it picks.
The router does filter by what the request needs. If your request involves image understanding, tool calling or structured outputs, it only considers models that support those features. Using the router itself is free of charge.
If you need one specific model, use that model's ID with the :free suffix instead, for example google/gemma-4-31b-it:free. That gives you repeatable behavior, which the router cannot.
Fallbacks and surprise charges
Used on its own, the router falls back only to other free models. The risk comes from fallback arrays you write yourself. If a paid model sits after a :free model in the array, a fallback to it is billed at that model's own price, even though the first model was free.
Keep paid models out of any array where you expect zero cost.
Top AI Credits for Startups
Apply directly through these verified programs.
OpenRouter Up to $3,000 + 0% fees for 12 months
Eligible for early-stage startups
Claude (via program) $100,000
Eligible for early-stage startups
OpenAI (via program) $2,500
Eligible for early-stage startups
Anthropic $20,000
Eligible for early-stage startups
AWS $300,000
Eligible for early-stage startups
Google Cloud $350,000
Eligible for early-stage startups
What is different about privacy on free models?
Training-on-prompts routing settings are separate for paid and free models, so you set each one on its own in your account settings.
OpenRouter lets you choose whether to allow routing to providers that may train on your data, according to their own policies. Because the settings are separate, a choice you made for paid models does not carry over to free ones. Check both.
The pricing table lists Data Policy-Based Routing as not included on the Free plan, while the same page's FAQ says Zero Data Retention routing is available on every plan, so check what your account settings offer. OpenRouter itself does not retain prompts unless you opt in to prompt logging. That covers OpenRouter, not the model providers behind each free model, whose own policies apply.
Do not send sensitive production data through free models until you have checked both settings.
What if the free limits are not enough?
You have two routes: buy at least $10 of credits to lift the daily cap to 1,000, or switch to the paid variant of the model, which has no platform-level request cap.
The paid route costs money per token, and prices differ by model. If cost is the concern, the cheapest LLM API ranking compares what you would actually pay.
The third route is free credits. Several model vendors and inference hosts offer credits you can spend on their own APIs, and OpenRouter itself has credits in the catalogue. These are the amounts tracked by AI Perks:
| Company | Free credits | How to get |
|---|---|---|
| OpenRouter | Up to $3,000 | Get this perk |
| Google Cloud (Vertex AI, Gemini) | $350,000 | Get this perk |
| Together AI | Up to $50,000 | Get this perk |
| Cerebras | Up to $22,500 | Get this perk |
Total across these four: $3,000 + $350,000 + $50,000 + $22,500 = $425,500 at the top of each range.
Eligibility and terms differ per company, and this guide does not list them. The catalogue holds the details. For more providers, the free AI API keys master guide covers the wider field.
Top AI Credits for Startups
Apply directly through these verified programs.
OpenRouter Up to $3,000 + 0% fees for 12 months
Eligible for early-stage startups
Claude (via program) $100,000
Eligible for early-stage startups
OpenAI (via program) $2,500
Eligible for early-stage startups
Anthropic $20,000
Eligible for early-stage startups
AWS $300,000
Eligible for early-stage startups
Google Cloud $350,000
Eligible for early-stage startups
How do you use OpenRouter free models step by step?
Start with credits, then pick between the router and a named model, and watch your daily count.
- Step 1: Get credits via AI Perks. Browse the catalogue at getaiperks.com for OpenRouter credits (up to $3,000) and for credits from the vendors behind the free models. Free credits carry you past the free tier onto paid models without paying list price yourself.
- Step 2: Check the current list. Open the models page filtered to free, since the list rotates. The table above is a snapshot from 2026-10-25.
- Step 3: Choose how to call it. Use
openrouter/freeif any free model will do. Use a model ID with:freeif you need a specific one. - Step 4: Handle errors carefully. Failed attempts count toward the 50-request daily cap, so add backoff instead of tight retries.
- Step 5: Keep fallbacks free. Do not put a paid model after a
:freeone unless you accept the charge. - Step 6: Raise the cap if needed. Buy at least $10 of credits for 1,000 requests per day, or move to the paid variant.
FAQ
How many free models does OpenRouter have?
On 2026-10-25 the models API listed 467 models in total and 20 of them were free, meaning zero prompt and completion price. That count moves as providers rotate models in and out. Check the models page for the live number, and see getaiperks.com for credits when the free tier runs short.
What is openrouter/free?
It is the Free Models Router. It picks one free model at random from those currently available, filtering for features your request needs such as vision, tool calling or structured outputs. It costs nothing to use and lists a 200,000-token context. You cannot choose the model it selects.
How do I pick one specific free model?
Add the :free suffix to that model's ID, such as nvidia/nemotron-3-ultra-550b-a55b:free. The router cannot be steered, so a named ID is the only way to get repeatable behavior. Some entries, like the Lyria models, are listed at zero price without the suffix. Credits for paid use are in the AI Perks catalogue.
Does making more accounts raise my free limit?
No. OpenRouter states that extra accounts or API keys do not affect your rate limits, because capacity is governed globally. The documented ways to raise the free daily cap are buying at least $10 of credits, which lifts it to 1,000, or switching to the paid variant of the model.
Do failed requests count against the free cap?
Yes. Failed attempts count toward the daily cap of 50 requests on free models, so a faulty retry loop can use your day quickly. A negative balance also blocks free models until you add credits. For funding options, the AI Perks catalogue lists current credit amounts.
Are free models private?
OpenRouter does not retain prompts unless you opt in to prompt logging. Training-on-prompts routing settings are separate for paid and free models, so review both. The Free plan shows no Data Policy-Based Routing in the pricing table. Providers behind each model apply their own data policies, so check them for anything sensitive.
Top AI Credits for Startups
Apply directly through these verified programs.
OpenRouter Up to $3,000 + 0% fees for 12 months
Eligible for early-stage startups
Claude (via program) $100,000
Eligible for early-stage startups
OpenAI (via program) $2,500
Eligible for early-stage startups
Anthropic $20,000
Eligible for early-stage startups
AWS $300,000
Eligible for early-stage startups
Google Cloud $350,000
Eligible for early-stage startups
Get past the free cap
The free list is a good place to test, but 50 requests per day ends a real project quickly. Credits from model vendors and inference hosts, plus OpenRouter credits up to $3,000, are the cheapest way past it. For a wider view of what is available, read the free AI API credits guide.
Get free AI credits at getaiperks.com →
Every amount and detail across the full catalogue of companies and perks is tracked at getaiperks.com.
