- Business
- EU
EU in-region routing for your LLM API calls
OpenRouter Business routes requests sent to eu.openrouter.ai only to shared provider endpoints in Europe (the EU/EEA, the UK, and Switzerland), where prompts and completions are processed. Account and billing data may be stored in the US. Requests served by your own BYOK key go to your deployment's region, which you control. Self-serve on Business: no contract, no monthly minimum.
8% platform fee on credit purchases, no monthly minimum, no contract.
Models available through eu.openrouter.ai
Read from the EU Models API and refreshed hourly. Prices are the EU endpoints' input and output token prices in US dollars per million tokens; other charges, such as images, caching, or web search, are on each model page.
- Anthropic: Claude Fable 5Via Google: input $11 output $55 per 1M tokens
- Anthropic: Claude Haiku 4.5Via Amazon Bedrock: input $1.10 output $5.50 per 1M tokensVia Google: input $1.10 output $5.50 per 1M tokens
- Anthropic: Claude Haiku 5.5Via Amazon Bedrock: input $0.11 output $0.55 per 1M tokensVia Google: input $0.11 output $0.55 per 1M tokens
- Anthropic: Claude Opus 4.5Via Amazon Bedrock: input $5.50 output $27.50 per 1M tokens
- Anthropic: Claude Opus 4.6Via Google: input $5.50 output $27.50 per 1M tokens
- Anthropic: Claude Opus 4.7Via Amazon Bedrock: input $5.50 output $27.50 per 1M tokensVia Google: input $5.50 output $27.50 per 1M tokens
- Anthropic: Claude Opus 4.8Via Amazon Bedrock: input $5.50 output $27.50 per 1M tokensVia Google: input $5.50 output $27.50 per 1M tokens
- Anthropic: Claude Opus 5Via Amazon Bedrock: input $5.50 output $27.50 per 1M tokensVia Google: input $5.50 output $27.50 per 1M tokens
- Anthropic: Claude Opus 5.5Via Amazon Bedrock: input $4.40 output $22 per 1M tokensVia Google: input $4.40 output $22 per 1M tokens
- Anthropic: Claude Sonnet 4.5Via Amazon Bedrock: input $3.30 output $16.50 per 1M tokens
- Anthropic: Claude Sonnet 4.6Via Amazon Bedrock: input $3.30 output $16.50 per 1M tokensVia Google: input $3.30 output $16.50 per 1M tokens
- Anthropic: Claude Sonnet 5Via Amazon Bedrock: input $2.20 output $11 per 1M tokensVia Google: input $2.20 output $11 per 1M tokens
- Anthropic: Claude Sonnet 5.5Via Google: input $2.20 output $11 per 1M tokens
- Amazon: Nova 2 LiteVia Amazon Bedrock: input $0.30 output $2.50 per 1M tokens
- Amazon: Nova Lite 1.0Via Amazon Bedrock: input $0.06 output $0.24 per 1M tokens
- Amazon: Nova Micro 1.0Via Amazon Bedrock: input $0.035 output $0.14 per 1M tokens
- Amazon: Nova Pro 1.0Via Amazon Bedrock: input $0.80 output $3.20 per 1M tokens
- DeepSeek: DeepSeek V4 Flash 0731Via Inceptron: input $0.13 output $0.5978 per 1M tokens
- Google: Gemini 2.5 FlashVia Google: input $0.30 output $2.50 per 1M tokens
- Google: Gemini 2.5 Flash LiteVia Google: input $0.10 output $0.40 per 1M tokens
- Google: Gemini 2.5 ProVia Google: input $1.25 output $10 per 1M tokens
- Google: Gemini 3.1 Flash LiteVia Google: input $0.275 output $1.65 per 1M tokens
- Google: Gemini 3.5 Flash LiteVia Google: input $0.33 output $2.75 per 1M tokens
- Mistral LargeVia Mistral: input $2.20 output $6.60 per 1M tokens
- Mistral Large 2407Via Mistral: input $2.20 output $6.60 per 1M tokens
- Mistral: Codestral 2508Via Mistral: input $0.33 output $0.99 per 1M tokens
- Mistral: Devstral 2 2512Via Mistral: input $0.40 output $2 per 1M tokens
- Mistral: Ministral 3 14B 2512Via Mistral: input $0.22 output $0.22 per 1M tokens
- Mistral: Ministral 3 3B 2512Via Mistral: input $0.11 output $0.11 per 1M tokens
- Mistral: Ministral 3 8B 2512Via Mistral: input $0.165 output $0.165 per 1M tokens
- Mistral: Mistral Large 3 2512Via Mistral: input $0.55 output $1.65 per 1M tokens
- Mistral: Mistral Large 4Via Mistral: input $0.748 output $2.299 per 1M tokens
- Mistral: Mistral Medium 3Via Mistral: input $0.44 output $2.20 per 1M tokens
- Mistral: Mistral Medium 3.1Via Mistral: input $0.44 output $2.20 per 1M tokens
- Mistral: Mistral Medium 3.5Via Mistral: input $1.65 output $8.25 per 1M tokens
- Mistral: Mistral NemoVia Mistral: input $0.15 output $0.15 per 1M tokens
- Mistral: Mistral Small 3.2 24BVia Mistral: input $0.10 output $0.30 per 1M tokens
- Mistral: Mistral Small 4Via Mistral: input $0.165 output $0.66 per 1M tokens
- Mistral: Mixtral 8x22B InstructVia Mistral: input $2.20 output $6.60 per 1M tokens
- Mistral: SabaVia Mistral: input $0.22 output $0.66 per 1M tokens
- Mistral: Voxtral Small 24B 2507Via Mistral: input $0.11 output $0.33 per 1M tokens
- MoonshotAI: Kimi K2.6Via Inceptron: input $0.4323 output $3.20 per 1M tokens
- MoonshotAI: Kimi K2.7 CodeVia Inceptron: input $0.6712 output $3.35 per 1M tokens
- OpenAI: GPT-4.1Via Azure: input $2.20 output $8.80 per 1M tokens
- OpenAI: GPT-4.1 MiniVia Azure: input $0.44 output $1.76 per 1M tokens
- OpenAI: GPT-4.1 NanoVia Azure: input $0.11 output $0.44 per 1M tokens
- OpenAI: GPT-4o-miniVia Azure: input $0.165 output $0.66 per 1M tokens
- OpenAI: GPT-5Via Azure: input $1.375 output $11 per 1M tokens
- OpenAI: GPT-5 MiniVia Azure: input $0.275 output $2.20 per 1M tokens
- OpenAI: GPT-5 NanoVia Azure: input $0.055 output $0.44 per 1M tokens
- OpenAI: GPT-5.1Via Azure: input $1.375 output $11 per 1M tokens
- OpenAI: GPT-5.4Via Azure: input $2.75 output $16.50 per 1M tokens
- OpenAI: GPT-5.5Via Azure: input $5.50 output $33 per 1M tokens
- OpenAI: GPT-5.6 LunaVia Azure: input $0.22 output $1.32 per 1M tokens
- OpenAI: GPT-5.6 Luna ProVia Azure: input $0.22 output $1.32 per 1M tokens
- OpenAI: GPT-5.6 SolVia Azure: input $4.40 output $22 per 1M tokens
- OpenAI: GPT-5.6 Sol ProVia Azure: input $4.40 output $22 per 1M tokens
- OpenAI: GPT-5.6 TerraVia Azure: input $2.20 output $13.20 per 1M tokens
- OpenAI: GPT-5.6 Terra ProVia Azure: input $2.20 output $13.20 per 1M tokens
- OpenAI: GPT-6 LunaVia Azure: input $0.11 output $0.55 per 1M tokens
- OpenAI: GPT-6 Luna ProVia Azure: input $0.11 output $0.55 per 1M tokens
- OpenAI: GPT-6 SolVia Azure: input $2.20 output $11 per 1M tokens
- OpenAI: GPT-6 Sol ProVia Azure: input $2.20 output $11 per 1M tokens
- OpenAI: GPT-6.1 SolVia Azure: input $2.20 output $11 per 1M tokens
- OpenAI: GPT-6.1 Sol ProVia Azure: input $2.20 output $11 per 1M tokens
- OpenAI: gpt-oss-120bVia Amazon Bedrock: input $0.15 output $0.60 per 1M tokens
- OpenAI: gpt-oss-20bVia Amazon Bedrock: input $0.07 output $0.15 per 1M tokens
- Z.ai: GLM 5.2Via Inceptron: input $0.69 output $2.52 per 1M tokensVia Mistral: input $1.54 output $4.84 per 1M tokens
- Z.ai: GLM 5.3Via Inceptron: input $0.55 output $1.505 per 1M tokensVia Mistral: input $1.54 output $4.84 per 1M tokens
- Z.ai: GLM 5.3 FlashVia Inceptron: input $0.10 output $0.8625 per 1M tokens
What is processed where
In-region routing covers API payloads sent to eu.openrouter.ai. It does not move account or operational records, and it cannot change where resources you control are hosted.
Processed in Europe (requests to eu.openrouter.ai)
- Prompts and completions, decrypted in the EU and routed only to shared provider endpoints located in Europe (the EU/EEA, the UK, and Switzerland). Requests served by your BYOK key go to your own deployment instead.
- Endpoint selection: only shared endpoints onboarded for the EU region are candidates, so a request with no eligible endpoint fails instead of leaving the region.
- Server tools that run entirely in-region, such as tool search, datetime, and apply patch, and native PDF parsing.
- Broadcast traces for EU requests, sent from infrastructure inside the region and never queued outside it.
May be stored or processed in the US
- Account data, billing records, and authentication data.
- Security logs, abuse-monitoring metadata, and other operational records.
- Batch API jobs, which run on US-based storage and queueing infrastructure and are only accepted on openrouter.ai.
Your responsibility
- Sending EU workloads to eu.openrouter.ai, and enforcing it with a data-region guardrail.
- BYOK: where your own Bedrock, Azure, or Vertex deployment runs, and each key's region settings, which differ by provider. Regional routing and guardrails do not verify or move resources in your account.
- Broadcast destinations: host your collector, Langfuse instance, or bucket in the EU if observability data must stay in the region.
- Your GDPR assessment, as the controller, of whether this setup meets your obligations.
Sources: the in-region routing docs for behavior, and Exhibit A of the Data Processing Agreement for contract scope.
One base URL changes
Same key, same account
There is no separate EU account or EU key. The API key you use on openrouter.ai works on eu.openrouter.ai once your organization is on Business.
Change the base URL
Point your client at https://eu.openrouter.ai/api/v1. The request is decrypted in the EU and only shared provider endpoints in Europe (the EU/EEA, the UK, and Switzerland) are eligible to serve it.
Fails closed
If no eligible endpoint is available for the model, the request errors instead of leaving the region. OpenRouter never silently falls back to a shared endpoint outside the region. Requests served by your BYOK key go to the deployment it is configured for, so keep it in the region you need.
Out-of-region features are rejected
Features whose infrastructure processes data outside the region return an error on eu.openrouter.ai: web search, web fetch, files, image generation, Fusion, shell, sandboxed bash, PDF engines other than native, the Batch API, and multi-model routers such as the Auto Router.
curl https://eu.openrouter.ai/api/v1/chat/completions \
-H "Authorization: Bearer <OPENROUTER_API_KEY>" \
-H "Content-Type: application/json" \
-d '{
"model": "anthropic/claude-sonnet-5",
"messages": [{"role": "user", "content": "Hello"}]
}'The full, current list of what regional domains reject is in Feature availability.
Require the EU endpoint with a data-region guardrail
The base URL is a per-request choice. A guardrail makes it a rule, at the workspace, member, or API-key level.
- 1
Workspace
Set Data regions to EU on the workspace default guardrail. Every request in the workspace must then arrive through eu.openrouter.ai.
- 2
Member
Assign a guardrail with Data regions set to EU to an organization member. It applies to all of their API keys and chatroom usage.
- 3
API key
Assign a guardrail directly to a key. Guardrails intersect, so a key or member guardrail can only narrow the regions the workspace allows.
curl -X PATCH https://openrouter.ai/api/v1/guardrails/GUARDRAIL_ID \
-H "Authorization: Bearer <OPENROUTER_MANAGEMENT_KEY>" \
-H "Content-Type: application/json" \
-d '{
"allowed_data_regions": ["europe"]
}'A request that arrives through a domain the guardrail does not allow is rejected with 403 Forbidden before any inference or endpoint selection. Data-region guardrails require a Business or Enterprise plan. The guardrail value for the EU is europe; the Models API filter is region=eu.
Setup in the editor and the API: Enforcing in-region routing with guardrails.
OpenRouter EU questions
Is OpenRouter US-only?
No. OpenRouter Business routes requests sent to eu.openrouter.ai only to shared provider endpoints in Europe (the EU/EEA, the UK, and Switzerland), and requests sent to us.openrouter.ai only to shared provider endpoints in the US. Requests sent to openrouter.ai use global routing. Requests served by your own BYOK key go to your cloud deployment, whose region you set.
Does EU routing need Enterprise?
No. EU in-region routing is self-serve on Business: no contract, no monthly minimum. An organization admin upgrades from settings, and Business adds a 8% platform fee on credit purchases.
OpenRouter is a US company: what about data transfers?
OpenRouter is a US company, and requests sent to eu.openrouter.ai are decrypted in the EU and routed only to shared provider endpoints located in Europe (the EU/EEA, the UK, and Switzerland), so their prompts and completions are processed there. Requests served by your own BYOK key go to your cloud deployment, whose region you set. Account data, billing records, authentication data, security logs, abuse-monitoring metadata, and other operational records may be stored and processed in the US. Transfers are covered by the DPA, which incorporates the EU Standard Contractual Clauses. The DPA includes an EU-only processing addendum (Exhibit A) for API payloads, which applies when you elect it in writing. Its scope is set by the DPA, not by the routing region.
Is OpenRouter GDPR compliant?
OpenRouter supports GDPR programs with a DPA and Standard Contractual Clauses. Whether your use of OpenRouter meets your GDPR obligations is your determination as the controller; EU in-region routing, data-region guardrails, and Zero Data Retention are the controls available to support it.
Which models are available on eu.openrouter.ai?
70 models from 5 providers are available through eu.openrouter.ai, including 13 Claude models. The table on this page is read hourly from the EU Models API (GET https://eu.openrouter.ai/api/v1/models) and leaves out :batch variants, which that API lists but regional domains reject.
What happens if no EU provider is available?
The request fails with an error; it is never routed to a shared endpoint outside the region. In-region routing fails closed: if no eligible endpoint is onboarded for the model, or a feature in the request would process data outside the region, OpenRouter returns an error instead of falling back. Requests served by your BYOK key go to the deployment it is configured for, and OpenRouter does not verify where it runs: provision it in the region you need and follow the provider-specific BYOK steps.
Does OpenRouter store my prompts?
Not for requests sent to eu.openrouter.ai: OpenRouter input/output logging is skipped for every regional endpoint request, even when it is enabled. Providers handle prompts under their own data policies; turn on Zero Data Retention to route only to endpoints that do not retain them.
Check every claim
- Data Processing Agreement
- In-region routing docs
- Sovereign AI docs
- Guardrails docs
- BYOK with in-region routing
- Business plan and pricing
Last reviewed on .
Upgrade from your settings
8% platform fee on credit purchases, no monthly minimum, no contract.