AI models and chat

Language models for chat, text, code and tool calling

LLM APIs give your code access to large language models for chat, text generation, summaries, translation, coding help and tool calling. Many follow the OpenAI request format, so one client library works with several providers. The free plans have rate or token limits, and some may use your prompts to improve their models, so read the terms before you send private data.

Bytez Model API

The Bytez Model API runs open models from Hugging Face, such as Qwen and Llama, and closed models from providers like OpenAI through one API. POST a JSON list of messages to /models/v2/, with the key in the ...

API, JSON, REST

Cloudflare Workers AI

Workers AI runs open AI models on Cloudflare's global network: text generation and chat, text embeddings, image generation, image classification, speech recognition, translation and summarisation. Call the models from a ...

API, JSON, MCP, REST

Cohere API

Cohere offers language AI for businesses: the Command models for chat and text generation, Embed for turning text into vectors for semantic search, and Rerank for ordering search results by relevance, all in more than 100 ...

API, JSON, REST

GitHub Models

GitHub Models lets developers try and compare AI models from OpenAI, Meta, Microsoft, Mistral, DeepSeek and others, in a playground on GitHub and through an API that uses a GitHub personal access token. It is meant for ...

API, JSON, REST

Google Gemini API

The Gemini API from Google gives access to the Gemini family of AI models for text generation, chat, summarising, coding, image and document understanding, speech, embeddings and function calling. Call it over REST or with the ...

API, JSON, REST, WebSocket

Groq API

Groq runs open AI models on its own LPU chips, which makes responses very fast. The API offers chat and text generation with open models (such as Llama, Qwen and GPT-OSS), speech-to-text with Whisper and text-to-speech, and is ...

API, JSON, MCP, REST

Kilo AI Gateway

The Kilo AI Gateway gives one OpenAI-compatible endpoint for several hundred language models from many providers, so an application can switch models without code changes. POST JSON to /api/v1/chat/completions; the model ...

API, JSON, REST

OpenRouter

OpenRouter gives one API, compatible with the OpenAI format, for hundreds of AI models from many providers, including OpenAI, Anthropic, Google, Meta, Mistral and DeepSeek, with automatic fallbacks between providers. Several ...

API, JSON, MCP, REST

OVHcloud AI Endpoints

OVHcloud AI Endpoints serves open-weight models such as Llama 3.3 70B, Mistral Small, Qwen and gpt-oss through an OpenAI-compatible API, hosted in Europe, for example to add chat, summaries or image description to an application. ...

API, JSON, REST

Requesty

Requesty is an LLM gateway with one OpenAI-compatible and Anthropic-compatible endpoint for several hundred models, with routing, caching and failover between providers. POST JSON to /v1/chat/completions with the key as a Bearer ...

API, JSON, REST

Vercel AI Gateway

Vercel AI Gateway gives one OpenAI-compatible and Anthropic-compatible endpoint for hundreds of models from OpenAI, Google, Anthropic, Meta, Mistral and others, with fallback between providers. POST JSON to /v1/chat/completions ...

API, JSON, REST

Z.ai API (GLM Flash models)

The Z.ai API gives access to the GLM language models from Zhipu AI for chat, text generation, coding and image understanding. It uses the OpenAI format, so the OpenAI SDKs work after changing the base URL to api.z.ai. Requests ...

API, JSON, REST

Compare the LLM and AI model APIs

ServiceAPI keyFormatsFree use
Bytez Model API Needed, free JSON Free with a free API key and $1 of credit that refreshes every 4 weeks, for open models up to 7B parameters (one request at a time) and closed models; paid plans unlock larger models.
Cloudflare Workers AI Needed, free JSON The Workers Free plan includes 10,000 'neurons' (Cloudflare's unit of AI compute) per day; requests above that fail until the daily reset, unless you move to the paid plan.
Cohere API Needed, free JSON Free trial key without a time limit, limited to 1,000 calls per month and a per-minute rate limit; production use needs a paid key.
GitHub Models Needed, free JSON Free for every GitHub account, with daily and per-minute request limits per model that depend on your Copilot plan; paid usage is available for more.
Google Gemini API Needed, free JSON Free tier for most text, speech and embedding models with a free API key and rate limits; content sent on the free tier may be used by Google to improve its products, so use the paid tier for private data.
Groq API Needed, free JSON Free plan with a free API key and rate limits per model; paid plans for higher limits.
Kilo AI Gateway Not needed JSON Free without a key for models tagged :free, up to 200 requests per hour per IP address; other models need a Kilo API key and paid credits.
OpenRouter Needed, free JSON Free models are limited to 20 requests per minute and 50 per day, or 1,000 per day once you have bought at least 10 credits; other models are paid per token.
OVHcloud AI Endpoints Not needed JSON Free without a key at up to 2 requests per minute per IP address and per model; an API key from a paid Public Cloud project raises this to 400 requests per minute, billed per token.
Requesty Needed, free JSON Free with a free API key for models marked free, up to 200 requests per day, no credit card; other models are paid per token.
Vercel AI Gateway Needed, free JSON Free with a free API key and a monthly included credit for a subset of models, at lower rate limits; Vercel may ask for a payment method before the free credit can be used.
Z.ai API (GLM Flash models) Needed, free JSON Free with a free API key for the Flash models: GLM-4.7-Flash, GLM-4.5-Flash and GLM-4.6V-Flash.

Questions

Which free LLM and AI model APIs work without an API key?

Kilo AI Gateway and OVHcloud AI Endpoints work without a key, according to their documentation. Most keyless services ask you to keep the number of requests reasonable, and some ask for a User-Agent that names your app.

Are these LLM and AI model APIs really free?

The services here can be used without paying, at least on a free plan or free tier, often with limits such as a number of requests per day or non-commercial use only. The table above quotes what each service says about its free use; check the provider's own terms before you rely on it.

How up to date is this list?

The address of every service is checked about once a week, and services that stop working or stop being free are removed. This list last changed on November 21st, 2026.