# Munito AI > Open-model inference provider for Europe, private by default. - OpenAI base URLs: `https://api.munito.ai/inference/v1` (global), `https://api.eu.munito.ai/inference/v1` (EU), `https://api.us.munito.ai/inference/v1` (US). - Anthropic base URLs: the same hosts without `/v1`, such as `https://api.munito.ai/inference`. - Regions: EU and US. A regional host processes requests in its region only. The global host can use either region, and the `munito-region` response header names the region. - Authentication: send the API key in `Authorization: Bearer ` or `X-API-Key: `. Create a key at https://console.munito.ai/keys. - Compatibility: the OpenAI Chat Completions, Responses and speech to text APIs, and the Anthropic Messages API. Keep your SDK, and change the base URL and the key. - Models and prices: https://docs.munito.ai/models. `GET /models` returns the public models in the OpenAI format. - Docs: each page has a Markdown twin at its URL plus `.md`. The full text is https://docs.munito.ai/llms-full.txt. ## Documentation - **Get started** - [Build with Munito AI](https://docs.munito.ai): Open models and document AI on European infrastructure. - [OpenAI- and Anthropic-compatible API for open models: send your first request](https://docs.munito.ai/quickstart): Call open models on Munito AI from an empty project in cURL, Python or TypeScript, through the OpenAI-compatible or the Anthropic-compatible API. - [OpenAI-compatible API for open models: change the base URL](https://docs.munito.ai/migrate-from-openai): Move from the OpenAI or Anthropic SDK to open models on Munito AI. Change the base URL, the API key and the model name, and check what works differently. - [OpenAI- and Anthropic-compatible API for open models: what each endpoint supports](https://docs.munito.ai/api-compatibility): Munito AI serves open-source models through the OpenAI Chat Completions and Responses APIs and the Anthropic Messages API. See which tool calls, structured outputs and reasoning controls each endpoint supports. - [FAQ: open-model inference API for Europe](https://docs.munito.ai/faq): Short answers about the Munito AI API. Where requests run, the EU-only host, the open models, the API formats, prices and data retention. - [Munito AI company facts: open-model inference provider for Europe](https://docs.munito.ai/company): Facts about Munito AI, an open-model inference provider for Europe from Paris, backed by Y Combinator. What it sells, its APIs and open models, its EU and global API hosts, and its data policy. - **Models and pricing** - All models - [Models and pricing](https://docs.munito.ai/models): Each open model with its context, its EU and US regions, and its price in USD for each service tier. - [DeepSeek-V4-Flash](https://docs.munito.ai/models/deepseek-ai-deepseek-v4-flash): Call DeepSeek-V4-Flash through an OpenAI- and Anthropic-compatible API, served in the EU: its model ID, context, prices and a code example. - [DeepSeek-V4.1-Flash-1017 (preview)](https://docs.munito.ai/models/deepseek-ai-deepseek-v4-1-flash-1017): Call DeepSeek-V4.1-Flash-1017 through an OpenAI- and Anthropic-compatible API, served on the global host: its model ID, context, prices and a code example. - [Gemma 4 12B](https://docs.munito.ai/models/google-gemma-4-12b): Call Gemma 4 12B through an OpenAI- and Anthropic-compatible API, served in the EU: its model ID, context, prices and a code example. - [GLM-5.3-Flash](https://docs.munito.ai/models/zai-org-glm-5-3-flash): Call GLM-5.3-Flash through an OpenAI- and Anthropic-compatible API, served in the EU: its model ID, context, prices and a code example. - [Kolibri 1](https://docs.munito.ai/models/aleph-alpha-kolibri-1): Call Kolibri 1 through an OpenAI- and Anthropic-compatible API, served in the EU: its model ID, context, prices and a code example. - [Qwen3.8-27B](https://docs.munito.ai/models/qwen-qwen3-8-27b): Call Qwen3.8-27B through an OpenAI- and Anthropic-compatible API, served in the EU and US: its model ID, context, prices and a code example. - [Cygnet (preview)](https://docs.munito.ai/models/blockbrain-cygnet): Answer yes or no, choice and score questions with Cygnet, served in the EU, through the Decisions API. - [pplx-decider-v1.2 (preview)](https://docs.munito.ai/models/perplexity-ai-pplx-decider-v1-2): Answer yes or no, choice and score questions with pplx-decider-v1.2, served in the EU, through the Decisions API. - [LightOnOCR-2-1B](https://docs.munito.ai/models/lightonai-lightonocr-2-1b): Convert PDFs and images to Markdown with LightOnOCR-2-1B, served in the EU and US: its model ID, prices and a code example. - **Capabilities** - [Structured outputs with open models on an OpenAI-compatible API](https://docs.munito.ai/structured-outputs): Get JSON that matches your JSON schema from open models, through the OpenAI Chat Completions and Responses APIs and the Anthropic Messages API. - [Tool calling with open models on an OpenAI-compatible API](https://docs.munito.ai/tool-calling): Function calling with open models. Define tools, control tool_choice, handle parallel calls and send the results back with the OpenAI or Anthropic SDK. - [Image and video input](https://docs.munito.ai/vision): Send images and video to a vision model by URL or as base64, and the limits of each request. - [Document OCR API hosted in Europe: PDFs, scans and handwriting to Markdown](https://docs.munito.ai/document-ai): Convert PDFs, scans and handwritten pages to Markdown with an OCR API in the EU or the US. Extract tables and invoice fields, pay for each page, and keep zero data retention under a GDPR DPA. - [Speech to text](https://docs.munito.ai/speech-to-text): Transcribe recordings with the OpenAI audio API, or send short audio clips in a chat request. - [Streaming](https://docs.munito.ai/streaming): Stream tokens as server-sent events on Chat Completions, Responses and Messages, read the usage, and cancel a stream. - [Reasoning models](https://docs.munito.ai/reasoning): The public chat models reason by default. Read the reasoning, control its effort, turn it off, and avoid an empty answer. - [Prompt caching](https://docs.munito.ai/prompt-caching): Munito AI caches repeated prompt prefixes automatically, with no cache-write fee. Read the cached tokens in the usage. - **Agents and tools** - [Agents and tools](https://docs.munito.ai/guides): Use Munito AI with the clients and development tools you already know. - [Use Claude Code with open models through an Anthropic-compatible endpoint](https://docs.munito.ai/claude-code): Run Claude Code with open models such as GLM-5.3-Flash. Point it at the Anthropic-compatible endpoint of Munito AI with your key and the model names. - [Run Codex CLI with open-weight models](https://docs.munito.ai/codex): Run the Codex CLI with open-weight models such as GLM-5.3-Flash. Add Munito AI as a model provider, and Codex calls the OpenAI-compatible Responses API. - [Use OpenCode with open models](https://docs.munito.ai/opencode): Use open models such as GLM-5.3-Flash in OpenCode. Connect Munito AI from the provider list, or add it as a custom OpenAI-compatible provider. - [Use Hermes Agent with open models](https://docs.munito.ai/hermes): Add Munito AI to Hermes Agent with one plugin, or as a custom OpenAI-compatible provider, and run open models such as GLM-5.3-Flash. - [Use pi with open models](https://docs.munito.ai/pi): Add Munito AI to the pi coding agent with one command, and run open models such as GLM-5.3-Flash on the global host or in the EU. - [Use OpenClaw with open models](https://docs.munito.ai/openclaw): Add Munito AI to OpenClaw with one plugin, and run open models such as GLM-5.3-Flash on the global host or in the EU. - [MCP server](https://docs.munito.ai/mcp): Let an AI assistant read your analytics, usage and API keys, and create API keys. - [iOS](https://docs.munito.ai/ios): Connect an iOS client with a custom OpenAI-compatible provider. - **Recipes** - [PDF to Markdown](https://docs.munito.ai/pdf-to-markdown): Convert a printed, scanned or handwritten PDF or image to Markdown with the Munito AI OCR API, in the EU or the US, by URL or as base64. - [Invoice OCR API: extract invoices to JSON](https://docs.munito.ai/invoice-extraction): Extract invoice data to structured JSON. Read the invoice PDF with the Munito AI OCR API, then get its fields from a chat model with a JSON schema. - [Retrieval-augmented generation](https://docs.munito.ai/rag): Read documents with Munito AI OCR, index them with a local keyword index, and answer questions with citations from a Munito AI chat model. - [LangChain](https://docs.munito.ai/langchain): Use Munito AI models in LangChain through ChatOpenAI or ChatAnthropic, with streaming, tools and structured output. - [LlamaIndex](https://docs.munito.ai/llamaindex): Use Munito AI models in LlamaIndex through OpenAILike, with chat, streaming, tools and structured output. - **Platform** - [Administration](https://docs.munito.ai/administration): Read your organization, members, API keys, usage, spend limits and models with an API key. - [Overflow](https://docs.munito.ai/overflow): Let another model or provider answer a chat request when capacity is full. - [Rate limits](https://docs.munito.ai/rate-limits): Model limits, dynamic limits, and retries. - [Service tiers](https://docs.munito.ai/service-tiers): Priority, standard and flex, and when to use each one. - [Regions](https://docs.munito.ai/regions): Keep processing in the EU or the US. - [Errors](https://docs.munito.ai/errors): Each HTTP status that the Munito AI API returns, the error body of each API, the exact messages, the cause, the fix, and whether to retry. - [Production checklist](https://docs.munito.ai/production-checklist): Retries, timeouts, background requests, service tiers, overflow, regions, spend limits, streaming and the reasoning trap, before you go live on Munito AI. - [Dedicated GPU endpoints for open models in Europe](https://docs.munito.ai/dedicated-endpoints): Reserved GPU capacity for one open model, for your organization only, in the EU, the US or on the global host. Munito AI sets it up on request, by contract. - [Committed-spend discounts for open-model inference](https://docs.munito.ai/committed-spend): Commit to an amount of open-model inference for volume rates, with a contract, an order form and a monthly invoice in USD, net 30. - **Trust** - [GDPR LLM API with EU data residency and a DPA](https://docs.munito.ai/gdpr-data-residency): Keep LLM prompts, outputs and stored data in the EU with an EU-only API host, and find the GDPR Data Processing Agreement and the sub-processors. - [Zero data retention LLM API, no training on your data](https://docs.munito.ai/data-privacy): The Munito AI API keeps no prompts or outputs by default and does not train models on your data. See what it stores only when you ask, and for how long. - [Security](https://docs.munito.ai/security): Encryption in transit, API key permissions, account checks, vulnerability reports and certification status. - **Compare** - [Scaleway vs Munito AI](https://docs.munito.ai/compare/scaleway): Scaleway Generative APIs and Munito AI side by side for open models in the EU: regions, APIs, OCR, decision models and prices. - [Mistral AI vs Munito AI](https://docs.munito.ai/compare/mistral): The Mistral AI API and Munito AI side by side for open models in the EU: regions, data retention, APIs, OCR, decision models and prices. - [OVHcloud vs Munito AI](https://docs.munito.ai/compare/ovhcloud): OVHcloud AI Endpoints and Munito AI side by side for open models in the EU: regions, APIs, OCR, decision models and prices. - [Blade vs Munito AI](https://docs.munito.ai/compare/blade): Blade and Munito AI side by side for open models in Europe: regions, APIs, OCR, decision models and prices. - [Together AI vs Munito AI](https://docs.munito.ai/compare/together-ai): Together AI and Munito AI side by side for open models: regions, data retention, APIs, OCR, decision models and prices. - [Fireworks AI vs Munito AI](https://docs.munito.ai/compare/fireworks-ai): Fireworks AI and Munito AI side by side for open models: regions, data retention, APIs, OCR, decision models and prices. - **Updates** - [Changelog](https://docs.munito.ai/changelog): Dated public changes to the Munito AI API, its models, regions, billing and docs. - [Model deprecations](https://docs.munito.ai/deprecations): Which public models Munito AI removed, when, and what a request for a removed model gets. - API reference - [API reference](https://docs.munito.ai/api): Request schemas, code examples, and interactive requests for the Munito AI API. - **Completions** - [List chat completions](https://docs.munito.ai/api/listChatCompletions): Lists the completions of your workspace that were created with store: true, oldest first by default. Page with after and limit. - [Chat completion](https://docs.munito.ai/api/createChatCompletion): Create a chat completion with an open model. Set the OpenAI SDK base URL to https://api.munito.ai/inference/v1 and use your Munito API key. - [Delete a chat completion](https://docs.munito.ai/api/deleteChatCompletion): Deletes a completion that was created with store: true, before the 30-day retention window ends. A deleted completion can no longer be retrieved. - [Retrieve a chat completion](https://docs.munito.ai/api/getChatCompletion): Returns a completion that was created with store: true or with Prefer: respond-async. A background completion answers 202 until it finishes. - [Update a chat completion](https://docs.munito.ai/api/updateChatCompletion): Replaces the metadata of a completion that was created with store: true. Metadata is the only field you can change. Send null to clear it. - [List chat completion messages](https://docs.munito.ai/api/listChatCompletionMessages): Lists the request messages of a completion that was created with store: true, in the order of the request by default. - **Messages** - [Create a message](https://docs.munito.ai/api/createMessage): Generate a response with the Anthropic Messages API. The Anthropic SDK uses your Munito key through X-API-Key. - [Count tokens](https://docs.munito.ai/api/countMessageTokens): Count the input tokens a Messages request would consume, without running it. Same request shape as POST /messages (minus max_tokens); free. - **Responses** - [Create a response](https://docs.munito.ai/api/createResponse): Create a response with the OpenAI Responses API. Send a string or a list of items in input. Set the system prompt with instructions. - [Retrieve a response](https://docs.munito.ai/api/getResponse): Returns a stored (or still-in-progress) response by id. - [Delete a response](https://docs.munito.ai/api/deleteResponse): Deletes a stored response and its items ahead of the 30-day retention window. A deleted response can no longer be retrieved or chained via previous_response_id. - [List input items](https://docs.munito.ai/api/listInputItems): Lists the input items of this response. The items of an earlier response in a previous_response_id chain stay with that response: list them from its own id. - [Cancel a background response](https://docs.munito.ai/api/cancelResponse): Cancels an in-progress response created with background: true. Returns the response object with status cancelled. - **Audio** - [Transcribe audio](https://docs.munito.ai/api/createTranscription): Turn speech into text with an audio model, in the request format of OpenAI's speech to text API. - **OCR** - [Run OCR](https://docs.munito.ai/api/createOcr): Document-AI workload: send a handwritten, scanned, or printed document (PDF or image) by URL or as base64, and get clean, structured Markdown back. - [Retrieve an OCR request](https://docs.munito.ai/api/getOcr): Returns the state of an OCR request: running with the pages that are done, then completed with the full result, or failed, cancelled or timed_out. - [Cancel an OCR request](https://docs.munito.ai/api/cancelOcr): Cancels an OCR request that is still running. - **Models** - [List models](https://docs.munito.ai/api/listModels): List public models without a key, in the OpenAI format: id, object, created, and owned_by. A valid key also includes models available to your account. - [Retrieve a model](https://docs.munito.ai/api/retrieveModel): Return one model from the list, in the same format. Public models need no key. A valid key also permits account-specific models. - **Administration** - [Authorization](https://docs.munito.ai/api/admin-authorization) - [Get an API key](https://docs.munito.ai/api/getApiKey): One API key of the organization. Its secret is never returned. - [Get spend limits](https://docs.munito.ai/api/getSpendLimits): Today's and this month's spend against the organization's spend limits. - [Get the organization](https://docs.munito.ai/api/getOrganization): The organization that the API key belongs to. - [Get usage](https://docs.munito.ai/api/getUsage): Requests, tokens, pages, errors and cost of a time range, with averages per request. - [List API keys](https://docs.munito.ai/api/listApiKeys): The API keys of the organization, the newest first. A key's secret is never returned. - [List available models](https://docs.munito.ai/api/listAvailableModels): The models that the organization can use, with their context length and inputs. - [List users](https://docs.munito.ai/api/listUsers): The members of the organization and their roles, the oldest first. - **Decisions** - [Evaluate decisions](https://docs.munito.ai/api/createDecision): Preview. Evaluate Noul, Choice, and Score questions against one state, with a TypeSafe-compatible response. - [Create a decision (OpenAI format)](https://docs.munito.ai/api/createOpenAIDecision): Preview. OpenAI's Decisions contract: ordered Predicate, Choice, and Score questions about one input, answered in the same order with probabilities.