EU data residency
Every model you can run in the EU through Opper, with where it runs, what the provider keeps and what it costs.
Each card shows only the model's routes in Europe, so the hosts, zero data retention and prices are the ones you get there. Norway applies the GDPR in full through the EEA, so its routes are included. For routes run only by European-owned providers, see sovereign AI models.
Claude Opus 5.5, Anthropic's model for long-running agentic coding, computer use, and knowledge work, with always-on adaptive thinking, a 1M-token context, and 128K output.
Claude Sonnet 5.5 is Anthropic's fast model for scoped coding, bug fixing, agent work, and polished documents, with adaptive thinking and a one million token context window.
Gemini 4 Argon is Google's flagship reasoning model for long, multi-step coding, security and knowledge work, with 1M context and text, image, audio, video and PDF input.
GPT-6.1 Sol combines near-Astra performance with lower cost for coding and professional workflows, with a 1.05M-token context and five reasoning effort levels.
Claude Opus 5 is Anthropic's flagship for complex agentic coding and enterprise work, with a 1M token context window and adaptive thinking.
GPT-5.6 Sol tops OpenAI's GPT-5.6 family, the frontier tier for complex coding, deep research, and long-running agents with a 1M token context.
GLM-5.3 by Z.ai is a 744B open-weight coding and agentic model, post-trained on the GLM-5.2 base with a 1M-token context and low, high, and max reasoning effort levels.
Kimi K3 by Moonshot: open-weight 2.8 trillion parameter MoE with 104 billion active, native vision, and a 1M token context for long-horizon coding and agentic work.
Claude Haiku 5.5 is Anthropic’s fast, low-cost model for high-volume summaries, classification, customer support, and scoped agent tasks.
GPT-5.6 Terra is the balanced middle tier of OpenAI's GPT-5.6 family, pairing near-flagship quality with everyday speed and a 1M token context.
Claude Opus 4.8 by Anthropic, highly capable model for complex reasoning, coding, and agentic work with 1M context and improved efficiency.
GLM-5.3-Flash by Z.ai is a natively multimodal 320B MoE with 18B active parameters, hybrid sparse and linear attention, a 1M-token context, and MIT open weights for coding and agent work.
Gemini 3.8 Flash is Google's workhorse multimodal model for software engineering and agentic tasks, pairing 1M context with image, audio, video, and PDF input.
Claude Opus 4.7 by Anthropic, advanced vision and coding model with 1M context for complex long-running agentic workflows.
Qwen3.8-2.4T-A95B is the open-weight release of Alibaba's Qwen3.8-Max flagship, a 2.4 trillion parameter Mixture-of-Experts activating 95B per token.
Qwen3.8 Flash-Next by Alibaba is a multimodal MoE with a 125B backbone and 6B active parameters, an early look at the Qwen4 architecture with 262K native context for coding and agents.
Gemini 3.7 Flash is Google's August 2026 Flash release, a large step up in coding and agentic automation over 3.6 while keeping the 1M-token multimodal context.
DeepSeek V4.1 Flash is a 552B multimodal MoE with a causal encoder-decoder design, 8B to 16B active parameters, a 1M-token context, and MIT weights for fast agentic coding.
DeepSeek V4.1 Flash (1017) is DeepSeek's October 2026 update to V4.1 Flash: the same multimodal MoE with a 1M-token context, now stronger at coding and agent work while writing shorter answers.
Mistral Large 4, Mistral AI's flagship open-weight model: a 1T-parameter mixture of experts (49B active) with native image input, tool use, optional reasoning and long context.
Claude Sonnet 5 is Anthropic's balanced production model, bringing near-Opus agentic coding and tool use to a 1M token context window.
GPT-6 Luna, OpenAI's fastest and most efficient GPT-6 model for focused, high-volume work like extraction, summarization, and classification, with a 1.05M-token context and six reasoning effort levels.
GPT-5.6 Luna, the fast high-volume tier of OpenAI's GPT-5.6 family, handles classification, summarization, and bulk pipelines with 1M context.
DeepSeek V4 Pro, flagship 1.6T MoE with 1M context and hybrid thinking for frontier reasoning and agents.
DeepSeek V4 Pro 0813 is the general availability build of DeepSeek's 1M-context flagship, tuned for agentic tool use and terminal-driven work.
DeepSeek V4 Flash 0731, the July 2026 retrained checkpoint with major agentic and coding gains over the original release.
DeepSeek V4 Flash, lightweight 284B MoE with 1M context and hybrid thinking for cost-efficient agents.
Gemini 3.6 Flash, Google's token-efficient multimodal Flash model, cuts output tokens by 17% while improving coding and agentic planning.
Qwen3.8 27B by Alibaba is a dense 27B open-weight vision-language model with image and video input, a 262K native context, thinking on by default, and Apache 2.0 weights.
GLM-5.2 by Z.ai is a 744B open-weight (MIT) mixture-of-experts coding model with a 1M-token context, IndexShare sparse attention, and selectable thinking-effort levels.
Gemini 3.5 Flash by Google pairs frontier-level intelligence with fast throughput and a 1M token context for agentic and coding tasks.
GPT-5.3 Codex is OpenAI's most capable agentic coding model, reaching state-of-the-art on SWE-Bench Pro and Terminal-Bench while using fewer tokens than prior models.
Claude Opus 4.6 by Anthropic, professional knowledge-work model with 1M context, extended thinking, and 128K token outputs.
Claude Sonnet 4.6 by Anthropic, full upgrade across coding, computer use, and long-context reasoning.
MiniMax M3: natively multimodal 428B Mixture-of-Experts model with ~23B active parameters, MiniMax Sparse Attention for long context, and 59.0% on SWE-Bench Pro
Claude Opus 4.5 by Anthropic, highly capable model for coding and agents with extended thinking and a tunable effort parameter.
Kimi K2.6 by Moonshot: open-weight agentic model with long-horizon coding, 300-agent swarms, video understanding, and 256K context.
Kimi K2.7 Code by Moonshot: open-weight, code-specialized agentic model on a 1 trillion parameter MoE, tuned for long-horizon software engineering and tool use.
GPT-5.4 Mini is OpenAI's strongest small model, improving over GPT-5 Mini in coding and reasoning while running more than 2x faster, with a 400k context window.
Kimi K2.5 by Moonshot: native multimodal 1T-parameter MoE with vision, 256K context, hybrid reasoning, and an agent swarm for parallel task execution.
MiniMax M2.5: 229B MoE model trained on hundreds of thousands of real environments, scoring 80.2% on SWE-Bench Verified and 76.3% on BrowseComp
Gemini 3.5 Flash Lite is Google's low-latency workhorse for high-volume automation, pairing 1M context with audio, video, and PDF understanding.
Qwen3.5 397B A17B by Alibaba: large sparse MoE model with 397B total (17B active) parameters for advanced reasoning and coding.
Qwen3.6-27B by Alibaba: 27B open-weight multimodal model with strong agentic coding (77.2 SWE-bench Verified) and reasoning across 262K context.
Claude Sonnet 4.5 by Anthropic, strong coding and agentic model scoring 61.4% on the OSWorld computer-use benchmark, 30+ hour focus.
GPT-5.4 Nano is the fastest, lightest GPT-5.4 variant, built for classification, data extraction, and lightweight coding subagents that need low latency.
GPT-5 Mini is a fast, cost-efficient GPT-5 model for well-defined tasks and precise prompts, delivering near-frontier intelligence for high-volume workloads.
GPT-5.1 Codex Mini by OpenAI, fast, cost-efficient code model with a 272K context for everyday development tasks.
Qwen3.6 35B-A3B by Alibaba: sparse MoE (35B total, 3B active) with vision, reasoning, and agentic coding for repository-level tasks.
Qwen3.6 35B A3B FP8 by Alibaba: fine-grained FP8 quantized variant of the sparse MoE model with near-identical performance.
Qwen3.5-122B-A10B by Alibaba: sparse MoE multimodal model with 122B total (10B active) parameters and 262K context for reasoning, vision, and agentic tasks.
Claude Haiku 4.5 by Anthropic, fast model matching Claude Sonnet 4 coding performance at a fraction of the cost, more than twice the speed.
Gemini 2.5 Pro by Google, an advanced reasoning LLM with 1M context and thinking mode, strong at complex coding, math, STEM, and deep document analysis.
GLM 4.7 Flash by Zhipu: 30B MoE (3B active) with 200K context, tuned for local deployment and fast agentic coding.
Gemma 4 31B is Google DeepMind's 30.7B parameter dense open model, the largest of the Gemma 4 family, with 256K context, thinking mode, and native function calling.
Mistral Medium 3.5 from Mistral — text model on the Opper gateway.
Qwen3.5 9B by Alibaba: 9B multimodal model with vision, tool calling, and reasoning for balanced performance and efficiency.
Gemini 2.5 Flash (Google). Fast multimodal model for text, images, video, and audio, with reasoning, tools, and structured output.
GPT-5 Nano is the fastest, lightest GPT-5 variant, built for summarization, classification, and simple tasks that need minimal latency.
Qwen3 235B A22B Instruct 2507 by Alibaba, an instruction-tuned MoE model with a 262K context window.
GPT-OSS 120B by OpenAI: open-weight Apache 2.0 reasoning model, 117B params (5.1B active), near o4-mini on reasoning, tools and structured output.
Mistral Small 4 (March 2026) unifies reasoning, coding, and multimodal in one MoE model with 256K context and configurable reasoning effort.
GPT-OSS 20B by OpenAI: open-weight Apache 2.0 model, 21B params (3.6B active), near o3-mini, for efficient and on-device reasoning.
Gemini 3.1 Flash Lite by Google, a cost-effective, low-latency multimodal LLM with 1M context, thinking mode, and tool use for efficient task automation.
Gemini 2.5 Flash Lite by Google, an efficient multimodal LLM with 1M context, optimized for cost and speed across text, vision, and structured tasks.
Qwen3 VL 30B A3B Instruct by Alibaba, an efficient open-weight MoE vision-language model with 3B active parameters and 131K context.
Llama 3.3 70B Instruct: refined 70B model with improved multilingual performance, structured output, and tool calling for reasoning workloads.
Llama 3.3 70B Instruct in an FP8 build, Meta's multilingual open-weight workhorse with 128K context, quantized for faster serving at near-identical quality.
Amazon Nova Pro, a balanced multimodal model combining accuracy, speed and cost for text, image and video understanding.
Amazon Nova Lite, a cost-effective multimodal model with 300K context for document analysis, visual Q&A, and video understanding.
Amazon Nova Micro, a fast text-only model for real-time summarization, translation, classification and simple reasoning.
GPT-5.3 Chat by OpenAI, chat-tuned model with a 272K context, vision, and tools for conversational use.
OpenAI's most capable embedding model, producing 3072-dimensional vectors with optional dimension reduction for high-accuracy multilingual retrieval and semantic search.
OpenAI's 1.55B-parameter speech-to-text model supporting 99 languages with 10-20% error reduction over Whisper Large v2, released November 2023.
OpenAI's pruned 809M-parameter speech-to-text model with 4 decoder layers, delivering much faster multilingual transcription at minimal quality loss.
OpenAI's legacy embedding model from December 2022, producing 1536-dimensional vectors, now superseded by the text-embedding-3 family.
Google Gemma 4 26B-A4B IT, mixture-of-experts open model with only 3.8B active parameters, multimodal reasoning, and function calling.
Google Gemma 3 12B IT, open-weight model with 128K context that handles text and image inputs across 140+ languages.
Google Gemma 3 27B IT, dense multimodal open model with 27.4B parameters for complex reasoning and vision tasks.
Gemini 2.5 Flash Image (Google). Fast multimodal model with native image generation and conversational editing.
Gemini 2.5 Flash TTS by Google, fast and cost-efficient text-to-speech with fine-grained voice control for real-time assistants and high-volume narration.
Gemini 2.5 Pro TTS by Google, premium text-to-speech for long-form content, professional narration, and creative workflows with high vocal clarity.
Gemini Embedding 2 by Google creates multimodal embeddings up to 3,072 dimensions for semantic search and retrieval.
Google Gemma 4 31B IT, instruction-tuned 31B dense model with multimodal reasoning, function calling, and 256K context.
Mistral OCR 4.1, the current model behind Mistral's Document AI stack, adding paragraph-level bounding boxes and block-level confidence scores to extracted text.
Mistral Medium 3.5, dense 128B open-weight model with 256k context optimized for agentic and coding workflows.
Mistral Small 3.2 24B Instruct is a 128K-context open-weight text and tool-calling model tuned for reliable instruction following and lean deployments.
Mistral Small 3.1 24B is an open-weight instruction-tuned model with 128K context, vision, and function calling at around 150 tokens per second.
Mistral Ministral 3 14B: largest Ministral 3 model, 256k context, Apache 2.0, tuned for local and edge deployment.
Mistral Ministral 3 3B: smallest Ministral 3 model, 256k context, Apache 2.0, ultra-efficient edge deployment.
Mistral Ministral 3 8B, edge-optimized multimodal model with 256k context, vision, and structured outputs.
Mistral Large 3, open-weight mixture-of-experts model with 256k context and native multimodal vision.
Mistral Codestral 25.08: code-optimized model for fill-in-the-middle completion and fast code generation, 128k context.
Pixtral Large is a 124B multimodal vision model with 128K context, built for document analysis, visual reasoning, and frontier image understanding.
Voxtral Mini Latest routes to the current Voxtral Mini, a 32K-context speech model for voice agents with native audio understanding.
Voxtral Mini Transcribe Realtime 2602 delivers streaming live transcription with 13-language support, speaker diarization, and a 4B edge-deployable model.
Voxtral Mini Transcribe Realtime by Mistral: streaming speech-to-text with configurable delay for live transcription, voice agents, and real-time subtitling.
Voxtral Mini Transcribe Realtime Latest routes to the current streaming transcription model, a 4B 13-language model for voice agents and live apps.
Voxtral Small 2507 by Mistral: 24B audio-native LLM with function calling and structured outputs for speech understanding and agentic chat.
Voxtral Small (latest) by Mistral: open-weight 24B audio-input instruct model that tracks the newest snapshot, with function calling and structured outputs.
Mistral OCR 3, document understanding model with a 74% win rate over OCR 2 on forms, tables, and handwritten text.
Voxtral Mini 2602 is a 4B streaming speech-to-text model for live applications, with speaker diarization and 13-language support.
Mistral Codestral Embed: embedding model specialized for semantic code search and retrieval, 8k context.
Mistral Codestral Embed 25.05: code retrieval embedding model Mistral reports beating Voyage Code 3, Cohere, and OpenAI.
Mistral Embed, general-purpose text embedding model producing 1024-dimensional vectors for semantic search and retrieval.
Mistral Embed 2312, efficient text embedding model with an 8k context for RAG and semantic search applications.
Voxtral Mini TTS by Mistral: 4B text-to-speech with zero-shot voice cloning from about 3 seconds of reference audio, supporting 9 languages.
Voxtral Mini TTS (latest) by Mistral: text-to-speech with zero-shot voice cloning and multilingual support, tracking the newest model snapshot.
Qwen3-VL-Plus, Alibaba's flagship vision-language model for document understanding, video analysis, and visual agents, with 262K context and hybrid thinking.
Qwen3-Embedding 8B by Alibaba, a 4096-dimensional embedding model with 36 layers and 32K context for dense retrieval and retrieval-augmented generation.
Qwen3 14B, Alibaba's April 2025 dense model with switchable thinking and direct modes for reasoning and tool use at 40K context.
Qwen3 30B A3B by Alibaba, an efficient MoE model with 3B active parameters and switchable thinking and non-thinking modes.
Qwen3 32B by Alibaba, a dense 32B model with switchable thinking and non-thinking modes for reasoning and speed.
Qwen3 8B, Alibaba's compact April 2025 dense model that can think step by step or answer directly, a low-cost choice for reasoning, extraction and tool calls.
Multilingual E5 Large by Microsoft: semantic search embedding model supporting 100 languages with 1024-dimensional representations.
Multilingual E5 Large Instruct by Microsoft: instruction-tuned embedding model with 1024 dimensions for multilingual semantic retrieval across 100 languages.
Azure Neural HD TTS is Microsoft's high-definition text-to-speech in Azure Speech, with context-aware voices that detect emotion in the input text and adjust tone in real time for natural, conversational speech.
Azure Neural TTS by Microsoft is the standard neural text-to-speech voice set of Azure Speech, with natural out-of-the-box voices in 100+ languages and locales and SSML control over pitch, pauses, pronunciation, speaking rate and style.
Amazon Nova 2 Lite, a fast multimodal model with extended thinking, 1M context, web grounding and code execution for reasoning tasks.
Cohere Rerank v3.5 is a multilingual semantic ranking model that reorders search and RAG results with state-of-the-art precision across 100+ languages and complex enterprise data.
BGE-reranker-v2-m3 by BAAI, lightweight 600M cross-encoder reranker built on BGE-m3 for fast multilingual relevance scoring.
IBM Docling: open-source OCR and document parsing with layout analysis, handling many formats including PDF, images, Office documents, and audio.
Apertus 70B, the Swiss AI Initiative's fully open model from EPFL and ETH Zurich, trained on 15T tokens with over 1,800 natively supported languages.
Brick Complexity Pro, Regolo's hosted prompt-complexity classifier that grades queries easy, medium, or hard so routing picks the right model tier.
Green L Raw, GreenPT's multilingual 24B chat and vision model based on Mistral Small 3.2, served without GreenPT's built-in system prompt so your own system prompt is the only one applied.
Green R Raw by GreenPT is the unprompted twin of its GPT-OSS-based Green R reasoning model: the same reasoning stack for analysis, writing and content generation, with no built-in GreenPT system prompt so your instructions are the only ones applied.
Green L, GreenPT's multilingual writing model built on Mistral Small 3.2 24B, with image understanding, Dutch grammar guardrails, and a 128K context.
Green R is GreenPT's branded reasoning model, built on OpenAI's open-weight GPT-OSS with GreenPT's sustainability system prompt, for analysis, writing and content generation over a 128K context.
Joint decisions over shared state: yes/no, choice and score questions, with probabilities for every answer.
Fish Speech S2.1 Pro, Fish Audio's multilingual text-to-speech and voice cloning model covering 83 languages from one model, with roughly 90ms time to first audio for live voice agents.
Fish Audio's S2.1 Pro text-to-speech model streamed in real time over WebSocket, turning text into speech as it arrives across 83 languages, for voice agents and live conversational apps.
Faster Whisper Large V3 is a multilingual speech-to-text model supporting 99 languages, a CTranslate2 reimplementation of OpenAI's Whisper that runs up to 4x faster at the same accuracy.
Eevo 1, Upright's consequential decision model, answers yes/no, choice and score questions about a case with a probability for every option, weighing what each option leads to for the people affected.
Sao10K's L3.3 70B Euryale v2.3, a 70B Llama 3.3 creative roleplay model and the direct successor to v2.2.
ALIA 40B Instruct 2601, Spain's publicly funded 40B multilingual model from the Barcelona Supercomputing Center, Apache 2.0 and pretrained across 35 European languages.
Clef, Cloudflare's 27B open-weight decision model, answers yes/no, choice and score questions about text or JSON with a probability for every option instead of writing text.
Clef-flash, Cloudflare's fast 9B open-weight decision model, answers yes/no, choice and score questions about text or JSON with a probability for every option in tens of milliseconds.
Undi95's ReMM SLERP L2 13B, a Llama 2 SLERP merge recreating the MythoMax recipe with updated component models.
Laya, ConvAI Innovations' open encoder decision model, answers yes/no, choice and score questions about short text with calibrated probabilities in a single forward pass.
KBLab's KB Whisper Large is a 2-billion-parameter Swedish speech-to-text model trained on 50,000 hours of audio, cutting word error rate by an average 47% versus Whisper-large-v3.
Norwegian and English speech-to-text from the National Library of Norway, a 1.55B-parameter Whisper model trained on 66,000 hours including parliamentary speeches and broadcasts.
Pianissimo, Klang's streaming speech recognition model for Swedish, transcribing live audio into punctuated text as people speak, dialects included.
Every route has its own model id. Call that id on any plan and the request is served on that route, in that location. The examples below call Claude Opus 5.5 on AWS Bedrock in the EU.
Copy this into a coding agent like Claude Code, Cursor or Codex and it will wire up Opper and call this route.
import OpenAI from "openai";const client = new OpenAI({apiKey: process.env.OPPER_API_KEY,baseURL: "https://api.opper.ai/v3/compat",});const completion = await client.chat.completions.create({model: "aws/claude-opus-5-5",messages: [{ role: "user", content: "Hello" }],});console.log(completion.choices[0].message.content);
A model access rule with Inference location: EU, set for your organization or a project, means no LLM or embedding call can reach a route outside the EU, even by mistake. Rules are part of the Control Plane. How model access rules work
For most teams EU hosting is the right setting. Inference runs in European data centres operated by AWS, Azure, Google and European hosts, Opper itself runs in AWS Stockholm, and many routes carry zero data retention as well.
Opper also runs as its own instance on evroc, the Swedish cloud, with routes to European-owned providers hosted in Europe.
This list as data. The same routes as JSON at opper.ai/models/eu.json (model id, provider, country, zero data retention, training and price per route) and as markdown at opper.ai/models/eu.md.
The whole catalogue. The model catalogue is public JSON at https://api.opper.ai/v3/models, no API key needed. Filter it with query parameters, for example ?type=llm&limit=3, and page through it with limit and offset. Use it when you can fetch a URL but cannot connect to an MCP server. Coding agents that use MCP can connect to https://api.opper.ai/mcp. Two of the server's tools need no account and no sign-in: list_models searches the model catalogue by name, type, provider or capability, and get_guide returns short setup guides. Everything that touches your account needs you to approve the connection in the browser first. About the MCP server
