Free limits for every provider
The table comes from the live catalog. It has one row for each provider. Each value is the highest limit of any free model from that provider. One model can have a lower limit than the value in the row. A dash means that the provider does not publish that limit.
| Provider | Free models | Max req/min | Max req/day | Max tokens/min | Tokens/month (one model) | Strongest free model |
|---|---|---|---|---|---|---|
| ElectronHub | 84 | - | - | - | $0.25/week shared recurring credit | Qwen3.8 Flash (ElectronHub)qwen3.8-flash |
| Cloudflare Workers AI | 39 | - | - | - | ~5-10M | Qwen 3.8 27B (Cloudflare Workers AI)@cf/qwen/qwen3.8-27b |
| Inferbase | 38 | - | - | - | 5,000 requests/month; 100 requests and $0.50 usage/day shared | DeepSeek V4 Flash (fp8) (Inferbase)deepseek-v4-flash:fp8 |
| AI Horde | 31 | - | - | - | Ongoing volunteer queue; availability varies | Gemma 4 31B Heretic (AI Horde)koboldcpp/gemma-4-31B-it-heretic |
| ACLIDE | 29 | - | - | - | €20/month shared internal usage credits · no card | Qwen3.8 Max (ACLIDE)qwen/qwen3.8-max:free |
| Google AI Studio | 21 | 15 | 1,000 | 250,000 | ~3M | Gemini 3.8 Flashgemini-3.8-flash |
| Api.Airforce | 20 | 1 | 1,000 | - | free · 1,000 req/day · 1 req/min | gpt-oss-20b (Api.Airforce)gpt-oss-20b |
| NVIDIA NIM | 19 | 40 | - | - | free · 40 RPM | GLM 5.3 Flash (NV)z-ai/glm-5.3-flash |
| Cohere | 17 | 20 | 33 | - | ~1-2M | Command A+ (05-2026)command-a-plus-05-2026 |
| DreamPrompting | 14 | 100 | 5,000 | - | free · 500K tokens/day · 5,000 req/day | Gemini 3.1 Flash-Litegoogle/gemini-3.1-flash-lite |
| Mistral | 12 | 2 | - | 500,000 | ~50-100M | Mistral Codemistral-code-latest |
| ModelScope | 11 | - | 200 | - | Daily login: 200 credits + 50 for linked Alibaba Cloud; 24h expiry; shared; noncommercial experience | GLM-5.2 (ModelScope)ZhipuAI/GLM-5.2 |
| OVH AI Endpoints | 11 | 2 | - | - | free · 2/min per IP | Qwen3.8 27B (OVH)Qwen3.8-27B |
| Waterfall | 11 | - | - | - | free · community rate limit | nex-n2.5-pro (Waterfall)nex-n2.5-pro-free |
| xKiro | 9 | - | - | - | free · 500K tokens/day (1M with Telegram) | SenseNova 6.8 Flash-Lite (xKiro)sensenova/sensenova-6.8-flash-lite |
| Kilo Gateway | 8 | - | - | - | free · 200/hr per IP (trial) | Dots3 Note Preview (Kilo)dots-studio/dots-3-note-preview:free |
| OpenRouter | 8 | 20 | 50 | - | ~6M | Dots3 Note Preview (free)dots-studio/dots-3-note-preview:free |
| Groq | 7 | 30 | 14,400 | 70,000 | ~6M | Qwen3.8 27B (Groq)qwen/qwen3.8-27b |
| Requesty | 7 | - | 200 | - | free · 200 rpd | Ling 3.1 Flash (Requesty)novita/ling-3.1-flash |
| Routeway | 7 | 5 | 200 | - | free · zero-priced route | Muse Glimmer 30B (Routeway free)muse-glimmer-30b:free |
| Speka | 7 | - | - | - | $1/month shared model-usage allowance | Step 5 Preview (Speka)stepfun-ai/step-5-preview |
| APINex | 6 | 5 | - | - | 10M weighted tokens/day shared across APINex free routes; 4x token multiplier; 5 requests/minute/account | GLM 5.3 (APINex Free)free/glm-5.3 |
| Gizmo | 6 | - | - | - | 1,000 free-model requests/month · shared account quota | Qwen3.8 27B (Gizmo)qwen/qwen3.8-27b:free |
| SEA-LION | 6 | 10 | - | - | free · 10 RPM | Qwen SEA-LION v4.5 27B (SEA-LION)aisingapore/Qwen-SEA-LION-v4.5-27B-IT |
| Septor Labs | 5 | - | - | - | free · shared daily allowance | MiniMax M2.5 (Septor Labs)minimax-m2.5-free |
| Agnes AI | 4 | - | - | - | free · promo $0/token | Agnes 3.0 Flash (Agnes AI)agnes-3.0-flash |
| Aion Labs | 4 | 15 | - | 20,000 | free · 20k tok/day | Aion 3.0aion-labs/aion-3.0 |
| BlockRun | 4 | - | - | - | Ongoing $0 models · rate-limited · no monthly cash grant | Muse Glimmer 30B (BlockRun)nvidia/muse-glimmer-30b |
| CLōD | 4 | - | - | - | free · shared daily allowance | Gemma 4 31B IT (CLōD)Gemma 4 31B IT |
| LLM7 | 4 | 40 | - | - | Free tier: 100,000 tokens/24h shared | gpt-oss:20b (LLM7)gpt-oss:20b |
| LLMTR | 4 | - | - | - | Ongoing zero-priced route | Qwen3.8 27B (LLMTR)qwen/qwen3.8-27b-free |
| Router9 | 4 | - | - | - | 10,000 credits/month (~$1), shared; account-anniversary reset | DeepSeek V4 Flash 0731 (Router9)deepseek/deepseek-v4-flash |
| BazaarLink | 3 | 10 | 50 | - | free · shared rate limit | DeepSeek V4 Flash (BazaarLink free)deepseek/deepseek-v4-flash-0731free:free |
| BlazeAPI | 3 | - | - | - | Free · 200,000 shared tokens/day; resets daily | GLM 5.3 Flash (BlazeAPI)free/glm-5.3-flash |
| Zhipu AI | 3 | - | - | - | ~30M | GLM-4.7 Flashglm-4.7-flash |
| AMD Radeon Cloud | 2 | - | - | - | free · recurring daily account allowance | Qwen3.8 Flash Next (AMD Radeon Cloud)Qwen3.8-Flash-Next |
| Logfare | 2 | - | - | - | free · fair use · traffic logged | Gemma 4 26B (Logfare)gemma-4-26b |
| Moondream | 2 | - | - | - | $5/month shared workspace usage credits | Moondream 3.1 9B A2Bmoondream3.1-9B-A2B |
| Plugsky | 2 | - | - | - | Ongoing Free plan; shared fair use | Plugsky Microplugsky-micro |
| SimpleLLM | 2 | - | - | - | Shared hourly/daily free-model quota | Gemma E4B Free (SimpleLLM)free-gemma-4-E4B |
| Speechify | 2 | - | - | - | - | Simba 3.2 (Speechify)simba-3.2 |
| UnoRouter | 2 | 1 | - | - | free · 1 req/min per model | GLM 5.3 (UnoRouter free)glm-5.3:free |
| Free.ai | 1 | - | - | - | - | Kokoro (Free.ai)kokoro |
| Pollinations | 1 | - | - | - | - | DreamShaper 8 LCM (Pollinations, legacy sana alias)sana |
| Typhoon | 1 | - | - | - | Ongoing free research API (shared rate limits) | Typhoon 2.5 30B A3B Instructtyphoon-v2.5-30b-a3b-instruct |
Live data from the FreeLLMAPI catalog (version 2026.11.19, checked 19 November 2026). Free limits change often. The catalog is checked every day.
The catalog has 487 free model endpoints from 45 providers today. The values change often. Providers add models, remove models and change quotas without notice.
How to read RPM, RPD, TPM and TPD
| Limit | Meaning | Usual window |
|---|---|---|
| RPM | Requests per minute | A rolling 60 seconds |
| RPD | Requests per day | One day. The reset time is different for each provider. |
| TPM | Tokens per minute. Input and output tokens can both count. | A rolling 60 seconds |
| TPD | Tokens per day | One day |
| Monthly budget | An estimate of the free tokens in one month | One calendar month |
The limit that stops you first is the one that matters. For example, Groq documents 30 RPM, 1,000 RPD, 8,000 TPM and 200,000 TPD for openai/gpt-oss-120b on its free plan. At 30 requests per minute you use the daily 1,000 requests in about 34 minutes. A request with 9,000 tokens fails at once, because it is larger than the 8,000 TPM limit.
- Scope: Google applies Gemini limits to each Google Cloud project, not to each key. Groq applies limits to each organization. A second key in the same project or organization does not add quota.
- Reset time: Google resets requests per day at midnight Pacific time. Cloudflare resets its daily allocation at 00:00 UTC.
- Shared pools: some providers count one quota across all free models. OpenRouter counts the free daily requests across every
:freemodel on the account. - Other units: Cloudflare Workers AI counts neurons, not tokens. The free allocation is 10,000 neurons per day.
Verified limits for the main providers
These values come from each provider's own documentation, checked on 20 November 2026. Use the live table above for model-level values.
| Provider | Free limit (official) | Scope |
|---|---|---|
| Google AI Studio (Gemini) | Different for each model. Google shows your active limits in AI Studio. | Per Google Cloud project |
| Groq | openai/gpt-oss-120b: 30 RPM, 1,000 RPD, 8,000 TPM, 200,000 TPD | Per organization |
| OpenRouter | :free models: 20 RPM. 50 requests per day, or 1,000 per day after you buy 10 credits. | Per account, all free models together |
| Cloudflare Workers AI | 10,000 neurons per day on Workers Free | Per account |
| OVHcloud AI Endpoints | Anonymous: 2 requests per minute. With a key: 400 requests per minute. | Anonymous: per IP and model. Key: per project and model. |
| NVIDIA API Catalog | NVIDIA does not publish the values. Your limit shows in your account on build.nvidia.com. | Per account |
| Cohere | Trial keys: 20 requests per minute on chat models, 1,000 API calls per month | Per trial key |
How to stack free tiers
One free tier is small. Many free tiers together give more requests per day. To stack them, get one key from each provider. Then send each request to a provider that still has quota.
FreeLLMAPI does this for you. It is an open-source gateway that runs on your computer. It keeps four counters (RPM, RPD, TPM, TPD) for each platform, model and key. Before it sends a request, it checks all four counters. It also checks pools that cover a whole account, for example the OpenRouter daily cap and the Cloudflare neuron cap.
Install and send one request
curl -fsSL https://freellmapi.co/install.sh | bash
curl http://localhost:3001/v1/chat/completions \
-H "Authorization: Bearer <unified-key>" \
-H "Content-Type: application/json" \
-d '{"model": "auto", "messages": [{"role": "user", "content": "hi"}]}'Add your provider keys on the Keys page of the dashboard at http://localhost:3001. Copy the unified key from the same page. The response header X-Routed-Via shows the provider and the model that answered.
What happens at the limit
Most providers send HTTP 429 when you go over a limit. Some providers send a Retry-After header. FreeLLMAPI uses these signals in this sequence:
- A short 429 on a minute limit puts the key on a 90 second cooldown. The router tries the next model in the chain.
- A full daily quota puts the model and key on a longer cooldown: 2 minutes, then 10 minutes, then 1 hour, then 24 hours.
- If the provider gives a
Retry-Aftervalue that is longer, the router uses the provider value. - If an error body states the real limit, the router saves it. The next pre-check then stops the request before the provider rejects it.
When every key is on cooldown, the gateway returns 429. Free tiers have no SLA. The strongest free models often have the smallest daily caps, so the answer quality can go down late in the day.
Ways to get more requests per day
- Add keys from more providers. Each provider adds its own quota.
- Use small, fast models for simple tasks. Keep the large models with small daily caps for hard tasks.
- Send fewer tokens. A shorter prompt helps with TPM and TPD limits.
- Do not make many accounts with one provider to get more quota. Read the terms of each provider first. Many terms do not allow this.
- Read free LLM API commercial use before you use free tiers in a product.
Frequently asked questions
What are the rate limits of the free Gemini API?
Google sets different limits for each model and applies them to each Google Cloud project. Your active limits show in AI Studio. Requests per day reset at midnight Pacific time. The live table on this page shows the current catalog values.
What is the Groq free tier rate limit?
Groq lists limits for each model. For openai/gpt-oss-120b the free plan allows 30 requests per minute, 1,000 requests per day, 8,000 tokens per minute and 200,000 tokens per day. Limits apply to the whole organization.
Does a second API key double my free quota?
Usually no. Google counts limits per project and Groq counts them per organization. A new key in the same project or organization shares the same quota.
Which free LLM API has the highest rate limit?
It depends on the limit type. Some providers give many requests per minute but few per day. Sort the live table by the limit that stops your workload first, then stack several providers.
What does error 429 mean on a free LLM API?
It means that you went over a rate limit or a quota. Wait for the reset time, send fewer tokens, or send the request to a different provider.
Related guides
- Gemini API quota exceeded (429 RESOURCE_EXHAUSTED)
- Groq rate limit reached (429)
- NVIDIA NIM API rate limit
- Free LLM API with no credit card
- best free llm apis 2026
- unlimited free ai tokens
Sources
- Google AI Studio: Rate limits (checked 20 November 2026)
- GroqCloud: Rate limits (checked 20 November 2026)
- OpenRouter: API rate limits (checked 20 November 2026)
- Cloudflare Workers AI: Pricing (checked 20 November 2026)
- OVHcloud: AI Endpoints getting started (checked 20 November 2026)
- Cohere: Rate limits (checked 20 November 2026)
- NVIDIA developer forum: credits replaced by rate limits (checked 20 November 2026)
- FreeLLMAPI docs: Quota and cooldown engine (checked 20 November 2026)