Anonymous LLM APIs at a glance
| Provider | Anonymous limit | Base URL | Notes |
|---|---|---|---|
| OVHcloud AI Endpoints | 2 requests per minute, for each IP and each model | https://oai.endpoints.kepler.ai.cloud.ovh.net/v1 | Hosted in the EU. OVHcloud says that data is not stored. |
| Kilo Gateway | 200 requests per hour, for each IP | https://ai-gateway.kilo.ai/api/v1 | Free models only. Some routes go to providers that log prompts. |
| AI Horde | No fixed rate. Anonymous users have the lowest queue priority. | https://oai.aihorde.net/v1 | Volunteers run the models. Use the key 0000000000. |
OVHcloud AI Endpoints
OVHcloud documents two limits. Anonymous use allows 2 requests per minute, for each IP address and each model. Use with an API key allows 400 requests per minute, for each Public Cloud project and each model. A key needs a Public Cloud project with a payment method. If you go over a limit, OVHcloud returns error 429.
Anonymous request to OVHcloud
curl https://oai.endpoints.kepler.ai.cloud.ovh.net/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{"model": "gpt-oss-120b", "messages": [{"role": "user", "content": "hi"}], "max_tokens": 300}'OVHcloud says that the service runs in the European Union. Its documentation says that data is not stored or shared during or after model use. The product page says that your data is never used to train its models.
| Model | Free on | Context | Free limits (best offer) | Tools | Vision |
|---|---|---|---|---|---|
| Qwen3.8 27B | OVH AI Endpoints | 262K | 2 req/min 2/min per IP | Yes | No |
| Qwen3.6 27B | OVH AI Endpoints | 131K | 2 req/min 2/min per IP | Yes | No |
| Qwen3.5 397B | OVH AI Endpoints | 262K | 2 req/min 2/min per IP | Yes | No |
| Qwen3.5 9B | OVH AI Endpoints | 262K | 2 req/min 2/min per IP | Yes | No |
| GPT-OSS 120B | OVH AI Endpoints | 131K | 2 req/min 2/min per IP | Yes | No |
| GPT-OSS 20B | OVH AI Endpoints | 131K | 2 req/min 2/min per IP | Yes | No |
| Llama 3.3 70B | OVH AI Endpoints | 131K | 2 req/min 2/min per IP | Yes | No |
| Mistral Small 3.2 24B | OVH AI Endpoints | 131K | 2 req/min 2/min per IP | Yes | No |
Showing 8 of 11 matching free models. See all models.
Live data from the FreeLLMAPI catalog (version 2026.11.19, checked 19 November 2026). Free limits change often. The catalog is checked every day.
Kilo Gateway
Kilo Gateway allows calls to free models without authentication. Free model ids contain :free. Kilo identifies anonymous requests by IP address and limits them to 200 requests per hour.
Read the privacy note before you send private data. Kilo's documentation says that Auto Free "may route your requests to providers that log prompts and outputs and use them to improve their services". For NVIDIA models, the note says that use is logged to improve NVIDIA products.
| Model | Free on | Context | Free limits (best offer) | Tools | Vision |
|---|---|---|---|---|---|
| Dots3 Note Preview | Kilo Gateway | 512K | 200/hr per IP | Yes | Yes |
| Nemotron 3.5 Super VL | Kilo Gateway | 1M | Ongoing keyless free access | No | No |
| Nemotron 3 Ultra 550B | Kilo Gateway | 1M | 200/hr per IP (trial) | Yes | No |
| Kilo Auto Free | Kilo Gateway | 256K | 200/hr per IP | Yes | No |
| Free Router | Kilo Gateway | 200K | 200/hr per IP | Yes | Yes |
| Nemotron 3.5 Lightning | Kilo Gateway | 1M | 200/hr per IP | Yes | No |
| North Mini Code | Kilo Gateway | 256K | 200/hr per IP | Yes | No |
| Nemotron 3 Nano Omni Reasoning | Kilo Gateway | 256K | 200/hr per IP (trial) | No | Yes |
Live data from the FreeLLMAPI catalog (version 2026.11.19, checked 19 November 2026). Free limits change often. The catalog is checked every day.
AI Horde
AI Horde is a free service run by a community. Volunteers run workers on their own GPUs. The OpenAI-compatible endpoint supports POST /v1/chat/completions and GET /v1/models. Send the anonymous key 0000000000 as the Bearer token.
Anonymous request to AI Horde
curl https://oai.aihorde.net/v1/chat/completions \
-H "Authorization: Bearer 0000000000" \
-H "Content-Type: application/json" \
-d '{"model": "<a model id from /v1/models>", "messages": [{"role": "user", "content": "Hello"}]}'- Speed: requests wait in a queue. Registered users get better priority than anonymous users. A request can expire before a worker serves it.
- Models: the list changes as workers go online and offline. The context window is set by the worker, often 4K to 8K tokens.
- Privacy: the FAQ says that a worker operator can technically change the code to see prompts. Workers do not see your user id or IP address. The FAQ tells you to treat each request as public.
Honest caveats
- Low limits: 2 requests per minute at OVHcloud is too slow for a coding agent. Use anonymous APIs for tests or as extra capacity.
- Shared IP: the limits count per IP address. On a shared network or a cloud server, other users can use your quota.
- No guarantee: anonymous access can stop at any time. No provider gives an SLA for it.
- Privacy differs: OVHcloud says that it does not store data. Kilo can route to providers that log prompts. AI Horde workers run on unknown computers.
- Terms still apply: anonymous use is still subject to each provider's terms of service.
Troubleshooting anonymous calls
| Problem | Cause | Fix |
|---|---|---|
| 429 after two OVHcloud requests | The anonymous limit is 2 requests per minute for each IP and model | Wait one minute, or send the next request to a different model |
| 429 on Kilo after many requests | The anonymous limit is 200 requests per hour for each IP | Wait for the next hour, or use a provider with a free key |
| AI Horde request never returns | No worker serves that model now, or the queue is long | Pick a model from GET /v1/models that has active workers |
Empty answer with finish reason length | A reasoning model used the whole output cap on hidden reasoning | Set max_tokens to 300 or more |
| Limits reached on the first call | Other users on the same public IP use the same quota | Run the calls from a network that you control |
An anonymous API is good for a quick test, a demo or a fallback. It is not good for an agent that sends many requests. For more throughput, get a free key from a provider that does not ask for a card. Read free LLM API with no credit card for the list.
Combine anonymous and keyed providers
FreeLLMAPI is an open-source gateway that runs on your computer. Its catalog has keyless rows for OVHcloud and Kilo, and an AI Horde provider. These rows work with no account. You can add free keys from Google, Groq and NVIDIA later. The router then uses the keyed providers first and the anonymous rows as extra capacity.
Install and call the local endpoint
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"}]}'The unified key is a key for your local gateway only. Get it from the Keys page at http://localhost:3001. Requests go from your computer to the upstream providers. The router counts each anonymous limit, so it does not send more than 2 requests per minute to one OVHcloud model.
Frequently asked questions
Is there an LLM API without an API key?
Yes. OVHcloud AI Endpoints, Kilo Gateway and AI Horde accept some requests with no account. The limits are low, for example 2 requests per minute per IP and model at OVHcloud.
Can I use an LLM API anonymously?
You can call these APIs without an account. The provider still sees your IP address, and limits count per IP. Kilo and AI Horde do not promise that prompts stay private.
What is the AI Horde anonymous API key?
The anonymous key is 0000000000. Send it as the Bearer token to https://oai.aihorde.net/v1. Anonymous requests have the lowest priority in the queue.
Is the OVHcloud AI Endpoints API free without an account?
Yes, for anonymous use at 2 requests per minute for each IP and model. Higher limits need an API key from a Public Cloud project with a payment method.
Related guides
- Free LLM API with no credit card
- Free OpenAI-compatible API base URLs
- OVH AI Endpoints free API
- Free LLM APIs hosted in Europe
- Do free LLM APIs train on your data?
Sources
- OVHcloud: AI Endpoints getting started (checked 20 November 2026)
- OVHcloud: AI Endpoints capabilities (checked 20 November 2026)
- OVHcloud: AI Endpoints product page (checked 20 November 2026)
- Kilo Gateway: Authentication (checked 20 November 2026)
- Kilo Gateway: Models and providers (checked 20 November 2026)
- AI Horde README (checked 20 November 2026)
- AI Horde FAQ (checked 20 November 2026)
- LLM7.io Quickstart (checked 20 November 2026)