Sovereign AI

Sovereign AI models hosted in Europe

Every model you can run with a European-owned provider in Europe, with where it runs, what the provider keeps and what it costs.

Sovereign AI on Opper
120 models207 routes23 providers11 home countries

Every model on a European-owned provider

Each card shows only the model's routes on European-owned providers, so the hosts, zero data retention and prices are the ones you get there. European-owned means the provider is headquartered in the EU, the EEA, Switzerland or the UK.

Moonshot’s Kimi K3.1 Pro is a million-token multimodal reasoning model for coding, agents and visual analysis.

Available on

Pricing from$2.00/ 1M input$10.00/ 1M output$0.20/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.30/ 1M input$1.20/ 1M output$0.0060/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

Open-weight, 1M-context MiniMax model for coding and agents with text, image and video input.

Available on

Pricing from$0.30/ 1M input$1.20/ 1M output$0.07/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

Xiaomi MiMo V2.6 Flash is a 309B parameter sparse MoE (15B active) omnimodal reasoning model tuned for efficient agentic coding and tool use, with a 1M token context.

Available on

Pricing from$0.40/ 1M input$1.21/ 1M output$0.16/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.77/ 1M input$2.31/ 1M output$0.08/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

DeepSeek V4 Pro, flagship 1.6T MoE with 1M context and hybrid thinking for frontier reasoning and agents.

Available on

Pricing from$2.00/ 1M input$4.00/ 1M output$0.50/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

DeepSeek V4 Pro 0813 is the general availability build of DeepSeek's 1M-context flagship, tuned for agentic tool use and terminal-driven work.

Available on

Pricing from$2.00/ 1M input$4.00/ 1M output$0.50/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

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

Available on

Pricing from$0.40/ 1M input$2.00/ 1M output$0.10/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$3.00/ 1M input$15.00/ 1M output$0.45/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$2.50/ 1M input$6.00/ 1M output$0.63/ 1M cache read262Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.20/ 1M input$0.50/ 1M output$0.05/ 1M cache read262Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.22/ 1M input$1.11/ 1M output$0.01/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

DeepSeek V4 Flash, lightweight 284B MoE with 1M context and hybrid thinking for cost-efficient agents.

Available on

Pricing from$0.11/ 1M input$0.28/ 1M output$0.02/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$1.22/ 1M input$4.39/ 1M output$0.19/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

Kimi K2.6 by Moonshot: open-weight agentic model with long-horizon coding, 300-agent swarms, video understanding, and 256K context.

Available on

Pricing from$0.80/ 1M input$3.75/ 1M output$0.21/ 1M cache read262Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.67/ 1M input$3.35/ 1M output$0.16/ 1M cache read262Kcontext

Privacy

ZDR
Training
No
Region
EU

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

Available on

Pricing from$0.36/ 1M input$1.46/ 1M output197Kcontext

Privacy

ZDR
Training
No
Region
EU

Qwen3.5 397B A17B by Alibaba: large sparse MoE model with 397B total (17B active) parameters for advanced reasoning and coding.

Available on

Pricing from$0.66/ 1M input$3.98/ 1M output262Kcontext

Privacy

ZDR
Training
No
Region
EU

Qwen3.6-27B by Alibaba: 27B open-weight multimodal model with strong agentic coding (77.2 SWE-bench Verified) and reasoning across 262K context.

Available on

Pricing from$0.15/ 1M input$1.35/ 1M output$0.07/ 1M cache read262Kcontext

Privacy

ZDR
Training
No
Region
EU

Qwen3.6 35B-A3B by Alibaba: sparse MoE (35B total, 3B active) with vision, reasoning, and agentic coding for repository-level tasks.

Available on

Pricing from$0.10/ 1M input$0.95/ 1M output$0.01/ 1M cache read262Kcontext

Privacy

ZDR
Training
No
Region
EU

Qwen3.6 35B A3B FP8 by Alibaba: fine-grained FP8 quantized variant of the sparse MoE model with near-identical performance.

Available on

Pricing from$0.33/ 1M input$1.33/ 1M output

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$1.11/ 1M input$4.64/ 1M output262Kcontext

Privacy

ZDR
Training
No
Region
EU

GLM 4.7 Flash by Zhipu: 30B MoE (3B active) with 200K context, tuned for local deployment and fast agentic coding.

Available on

Pricing from$0.07/ 1M input$0.41/ 1M output$0.01/ 1M cache read131Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.11/ 1M input$0.33/ 1M output$0.02/ 1M cache read256Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Medium 3.5 from Mistral — text model on the Opper gateway.

Available on

Pricing from$1.65/ 1M input$8.25/ 1M output262Kcontext

Privacy

ZDR
Training
No
Region
EU

Qwen3.5 9B by Alibaba: 9B multimodal model with vision, tool calling, and reasoning for balanced performance and efficiency.

Available on

Pricing from$0.04/ 1M input$0.07/ 1M output$0.02/ 1M cache read262Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.17/ 1M input$0.65/ 1M output131Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Small 4 (March 2026) unifies reasoning, coding, and multimodal in one MoE model with 256K context and configurable reasoning effort.

Available on

Pricing from$0.17/ 1M input$0.66/ 1M output$0.02/ 1M cache read256Kcontext

Privacy

ZDR
Training
No
Region
EU

GLM 4.5 Air by Zhipu: 106B compact MoE with 12B active parameters, tuned for efficient agentic workflows.

Available on

Pricing from$0.14/ 1M input$0.86/ 1M output$0.02/ 1M cache read131Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.0090/ 1M input$0.04/ 1M output$0.0030/ 1M cache read131Kcontext

Privacy

ZDR
Training
No
Region
EU

Qwen3 VL 30B A3B Instruct by Alibaba, an efficient open-weight MoE vision-language model with 3B active parameters and 131K context.

Available on

Pricing from$0.22/ 1M input$0.88/ 1M output131Kcontext

Privacy

ZDR
Training
No
Region
EU

Llama 3.3 70B Instruct: refined 70B model with improved multilingual performance, structured output, and tool calling for reasoning workloads.

Available on

Pricing from$0.99/ 1M input$0.99/ 1M output131Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$1.11/ 1M input$1.11/ 1M output128Kcontext

Privacy

ZDR
Training
No
Region
EU

Llama 3.1 8B Instruct: efficient open-weight small model with a 131K context, tool calling, and structured output for production deployments.

Available on

Pricing from$0.02/ 1M input$0.05/ 1M output$0.0030/ 1M cache read131Kcontext

Privacy

ZDR
Training
No
Region
EU

Google Gemma 4 31B IT, instruction-tuned 31B dense model with multimodal reasoning, function calling, and 256K context.

Available on

Pricing from$0.14/ 1M input$0.39/ 1M output$0.02/ 1M cache read262Kcontext

Privacy

ZDR
Training
No
Region
EU

OpenAI's 1.55B-parameter speech-to-text model supporting 99 languages with 10-20% error reduction over Whisper Large v2, released November 2023.

Available on

Pricing$0.0022 / min

Privacy

ZDR
Training
No
Region
EU

OpenAI's pruned 809M-parameter speech-to-text model with 4 decoder layers, delivering much faster multilingual transcription at minimal quality loss.

Available on

Pricing$0.0022 / min

Privacy

ZDR
Training
No
Region
EU

Google Gemma 3 12B IT, open-weight model with 128K context that handles text and image inputs across 140+ languages.

Available on

Pricing from$0.05/ 1M input$0.15/ 1M output131Kcontext

Privacy

ZDR
Training
No
Region
EU

Google Gemma 3 27B IT, dense multimodal open model with 27.4B parameters for complex reasoning and vision tasks.

Available on

Pricing from$0.08/ 1M input$0.30/ 1M output$0.04/ 1M cache read131Kcontext

Privacy

ZDR
Training
No
Region
EU

Google Gemma 4 26B-A4B IT, mixture-of-experts open model with only 3.8B active parameters, multimodal reasoning, and function calling.

Available on

Pricing from$0.07/ 1M input$0.24/ 1M output$0.01/ 1M cache read262Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral's 5M-parameter time-series foundation model: probabilistic forecasts of numeric series.

Available on

Pricing from$0.0050/ 1M input$0.05/ 1M output

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing$0.0044 / page

Privacy

ZDR
Training
No
Region
EU

Mistral Medium 3.5, dense 128B open-weight model with 256k context optimized for agentic and coding workflows.

Available on

Pricing from$1.66/ 1M input$6.63/ 1M output180Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.17/ 1M input$0.33/ 1M output131Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.36/ 1M input$0.36/ 1M output128Kcontext

Privacy

ZDR
Training
No
Region
EU

Voxtral Mini Transcribe 3 is Mistral's compact speech-to-text model for recorded audio, transcribing 50 languages with speaker diarization, segment timestamps and context biasing for names and terms.

Available on

Pricing$0.0039 / min

Privacy

ZDR
Training
No
Region
EU

Voxtral Mini Transcribe Realtime 3 is Mistral's streaming speech-to-text model for live audio and voice agents, with end-of-turn detection, speaker diarization and context biasing in one session.

Available on

Pricing$0.0088 / min

Privacy

ZDR
Training
No
Region
EU

Voxtral Small Transcribe 3 is Mistral's flagship speech-to-text model for recorded audio, its most accurate transcriber, with speaker diarization, segment timestamps and context biasing across 50 languages.

Available on

Pricing$0.0044 / min

Privacy

ZDR
Training
No
Region
EU

Voxtral TTS 3 is Mistral's text-to-speech model with expressive preset voices, inline emotion and pace tags, custom pronunciation and voice cloning, in 35 languages and more than 50 dialects.

Available on

Pricing$14.30 / 1M chars

Privacy

ZDR
Training
No
Region
EU

Mistral Ministral 3 14B: largest Ministral 3 model, 256k context, Apache 2.0, tuned for local and edge deployment.

Available on

Pricing from$0.22/ 1M input$0.22/ 1M output$0.02/ 1M cache read256Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Ministral 3 3B: smallest Ministral 3 model, 256k context, Apache 2.0, ultra-efficient edge deployment.

Available on

Pricing from$0.11/ 1M input$0.11/ 1M output$0.01/ 1M cache read256Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Ministral 3 8B, edge-optimized multimodal model with 256k context, vision, and structured outputs.

Available on

Pricing from$0.17/ 1M input$0.17/ 1M output$0.02/ 1M cache read256Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Large 3, open-weight mixture-of-experts model with 256k context and native multimodal vision.

Available on

Pricing from$0.55/ 1M input$1.65/ 1M output256Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Nemo Instruct, instruction-tuned 12B model with 128k context, multilingual strengths, and efficient quantized inference.

Available on

Pricing from$0.02/ 1M input$0.03/ 1M output$0.0034/ 1M cache read131Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Codestral 25.08: code-optimized model for fill-in-the-middle completion and fast code generation, 128k context.

Available on

Pricing from$0.33/ 1M input$0.99/ 1M output$0.03/ 1M cache read128Kcontext

Privacy

ZDR
Training
No
Region
EU

Voxtral Mini Transcribe Realtime by Mistral: streaming speech-to-text with configurable delay for live transcription, voice agents, and real-time subtitling.

Available on

Pricing$0.0066 / min

Privacy

ZDR
Training
No
Region
EU

Voxtral Small 2507 by Mistral: 24B audio-native LLM with function calling and structured outputs for speech understanding and agentic chat.

Available on

Pricing from$0.11/ 1M input$0.44/ 1M output33Kcontext

Privacy

ZDR
Training
No
Region
EU

Voxtral Small (latest) by Mistral: open-weight 24B audio-input instruct model that tracks the newest snapshot, with function calling and structured outputs.

Available on

Pricing from$0.11/ 1M input$0.44/ 1M output33Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral OCR 3, document understanding model with a 74% win rate over OCR 2 on forms, tables, and handwritten text.

Available on

Pricing$0.0022 / page

Privacy

ZDR
Training
No
Region
EU

Voxtral Mini 2602 is a 4B streaming speech-to-text model for live applications, with speaker diarization and 13-language support.

Available on

Pricing$0.0033 / min

Privacy

ZDR
Training
No
Region
EU

Voxtral Mini Latest is Mistral's alias for its current Voxtral Mini batch transcriber, Voxtral Mini Transcribe 2, turning recorded audio into text with speaker diarization, timestamps and context biasing.

Available on

Pricing$0.0033 / min

Privacy

ZDR
Training
No
Region
EU

Mistral Codestral Embed: embedding model specialized for semantic code search and retrieval, 8k context.

Available on

Pricing from$0.17/ 1M input$0/ 1M output$0.02/ 1M cache read8Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Codestral Embed 25.05: code retrieval embedding model Mistral reports beating Voyage Code 3, Cohere, and OpenAI.

Available on

Pricing from$0.17/ 1M input$0/ 1M output$0.02/ 1M cache read8Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Embed, general-purpose text embedding model producing 1024-dimensional vectors for semantic search and retrieval.

Available on

Pricing from$0.11/ 1M input$0/ 1M output$0.01/ 1M cache read8Kcontext

Privacy

ZDR
Training
No
Region
EU

Mistral Embed 2312, efficient text embedding model with an 8k context for RAG and semantic search applications.

Available on

Pricing from$0.11/ 1M input$0/ 1M output$0.01/ 1M cache read8Kcontext

Privacy

ZDR
Training
No
Region
EU

Voxtral Mini TTS by Mistral: 4B text-to-speech with zero-shot voice cloning from about 3 seconds of reference audio, supporting 9 languages.

Available on

Pricing$17.60 / 1M chars

Privacy

ZDR
Training
No
Region
EU

Voxtral Mini TTS (latest) by Mistral: text-to-speech with zero-shot voice cloning and multilingual support, tracking the newest model snapshot.

Available on

Pricing$14.30 / 1M chars

Privacy

ZDR
Training
No
Region
EU

Mistral's 123M-parameter time-series foundation model: probabilistic forecasts of numeric series.

Available on

Pricing from$0.03/ 1M input$0.30/ 1M output

Privacy

ZDR
Training
No
Region
EU

Mistral's 209M-parameter time-series foundation model: probabilistic forecasts of numeric series.

Available on

Pricing from$0.05/ 1M input$0.50/ 1M output

Privacy

ZDR
Training
No
Region
EU

Mistral's 28M-parameter time-series foundation model: probabilistic forecasts of numeric series.

Available on

Pricing from$0.01/ 1M input$0.10/ 1M output

Privacy

ZDR
Training
No
Region
EU

Qwen3-Embedding 8B by Alibaba, a 4096-dimensional embedding model with 36 layers and 32K context for dense retrieval and retrieval-augmented generation.

Available on

Pricing from$0.11/ 1M input$0/ 1M output

Privacy

ZDR
Training
No
Region
EU

Qwen3 14B, Alibaba's April 2025 dense model with switchable thinking and direct modes for reasoning and tool use at 40K context.

Available on

Pricing from$0.08/ 1M input$0.22/ 1M output$0.04/ 1M cache read41Kcontext

Privacy

ZDR
Training
No
Region
EU

Qwen3 30B A3B by Alibaba, an efficient MoE model with 3B active parameters and switchable thinking and non-thinking modes.

Available on

Pricing from$0.10/ 1M input$0.28/ 1M output$0.05/ 1M cache read41Kcontext

Privacy

ZDR
Training
No
Region
EU

Qwen3 32B by Alibaba, a dense 32B model with switchable thinking and non-thinking modes for reasoning and speed.

Available on

Pricing from$0.10/ 1M input$0.28/ 1M output$0.05/ 1M cache read41Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.45/ 1M input$0.45/ 1M output$0.22/ 1M cache read41Kcontext

Privacy

ZDR
Training
No
Region
EU

Multilingual E5 Large by Microsoft: semantic search embedding model supporting 100 languages with 1024-dimensional representations.

Available on

Pricing from$0.03/ 1M input$0/ 1M output

Privacy

ZDR
Training
No
Region
EU

Multilingual E5 Large Instruct by Microsoft: instruction-tuned embedding model with 1024 dimensions for multilingual semantic retrieval across 100 languages.

Available on

Pricing from$0.03/ 1M input$0/ 1M output

Privacy

ZDR
Training
No
Region
EU

BGE-m3 by BAAI, 569M multilingual embedding unifying dense, sparse, and multi-vector retrieval across 100+ languages and inputs up to 8192 tokens.

Available on

Pricing from$0.01/ 1M input$0/ 1M output8Kcontext

Privacy

ZDR
Training
No
Region
EU

BGE-reranker-v2-m3 by BAAI, lightweight 600M cross-encoder reranker built on BGE-m3 for fast multilingual relevance scoring.

Available on

Pricing from$0.11/ 1M input$0/ 1M output

Privacy

ZDR
Training
No
Region
EU

IBM Docling: open-source OCR and document parsing with layout analysis, handling many formats including PDF, images, Office documents, and audio.

Available on

Pricing$0.003 / page

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.44/ 1M input$2.32/ 1M output66Kcontext

Privacy

ZDR
Training
No
Region
EU

Brick Complexity Pro, Regolo's hosted prompt-complexity classifier that grades queries easy, medium, or hard so routing picks the right model tier.

Available on

Pricing from$0.11/ 1M input$0.44/ 1M output131Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.28/ 1M input$0.88/ 1M output131Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.39/ 1M input$1.05/ 1M output131Kcontext

Privacy

ZDR
Training
No
Region
EU

Green L, GreenPT's multilingual writing model built on Mistral Small 3.2 24B, with image understanding, Dutch grammar guardrails, and a 128K context.

Available on

Pricing from$0.28/ 1M input$0.88/ 1M output128Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.39/ 1M input$1.05/ 1M output123Kcontext

Privacy

ZDR
Training
No
Region
EU

Joint decisions over shared state: yes/no, choice and score questions, with probabilities for every answer.

Available on

Pricing from$0.13/ 1M input$0/ 1M output16Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing$0.0022 / min

Privacy

ZDR
Training
No
Region
EU

Reson8's speech-to-text model for prerecorded audio in ten European languages.

Available on

Pricingfrom $0.0004 / min

Privacy

ZDR
Training
No
Region
EU

Reson8's fast, low-cost speech-to-text for prerecorded audio, returning text only.

Available on

Pricing$0.0002 / min

Privacy

ZDR
Training
No
Region
EU

Reson8's live streaming speech-to-text, with end-of-turn detection for voice agents.

Available on

Pricingfrom $0.0004 / min

Privacy

ZDR
Training
No
Region
EU

REAX 1.3 is Decision Models' open typed-decision model: it answers one yes or no, choice or score question about a text with a probability for every option.

Available on

Pricing from$0.40/ 1M input$0/ 1M output29Kcontext

Privacy

ZDR
Training
No
Region
EU

Quasar 1.2 358B, Multiverse Computing's compressed version of Z.ai's GLM 5.2, a reasoning model for coding and agents.

Available on

Pricing from$0.11/ 1M input$0.33/ 1M output$0.0099/ 1M cache read203Kcontext

Privacy

ZDR
Training
No
Region
EU

Kolibri 1, Aleph Alpha's open-weight mixture-of-experts reasoning model for German and English, with 78B total and 3.46B active parameters, tool calling and long context.

Available on

Pricing from$0.16/ 1M input$0.56/ 1M output$0.09/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

Eevo-1 is Upright’s general-purpose SystemOne model, strongest at weighing consequences, stakes and expected value for tasks like model routing, coding-agent decision support and legal review.

Available on

Pricing from$0.10/ 1M input$0/ 1M output74Kcontext

Privacy

ZDR
Training
No
Region
EU

Blink v0.3 is Pixilab's fast decision model, a 26B mixture of experts with 4B active parameters built on Gemma 4 26B-A4B, for yes or no, choice and score answers.

Available on

Pricing from$0.03/ 1M input$0/ 1M output16Kcontext

Privacy

ZDR
Training
No
Region
EU

mammoth-1 is Base Labs' 27B open decision model, post-trained from Qwen3.5-27B, that answers yes or no, choice and score questions with calibrated probabilities.

Available on

Pricing from$0.10/ 1M input$0/ 1M output8Kcontext

Privacy

ZDR
Training
No
Region
EU

Heabsy's refusal-removed specialist model for cybersecurity, code analysis and security research.

Available on

Pricing from$1.21/ 1M input$3.64/ 1M output$0.22/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

dealignai's GLM 5.3 Flash derivative with weight-level refusal removal and a visual encoder.

Available on

Pricing from$0.15/ 1M input$0.59/ 1M output$0.05/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

A MiMo V2.6 Flash MOPD derivative with Heretic directional refusal ablation applied as a LoRA adapter.

Available on

Pricing from$0.40/ 1M input$1.21/ 1M output$0.22/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

A MiMo V2.6 Flash MOPD derivative carrying dealignai's weight-level refusal removal.

Available on

Pricing from$0.37/ 1M input$1.21/ 1M output$0.12/ 1M cache read1Mcontext

Privacy

ZDR
Training
No
Region
EU

huihui-ai's Qwen3.8 derivative with refusal directions removed and the visual encoder preserved.

Available on

Pricing from$0.42/ 1M input$1.81/ 1M output$0.06/ 1M cache read524Kcontext

Privacy

ZDR
Training
No
Region
EU

TeichAI's Qwen3.8 27B fine-tune distilled from assistant and coding conversations.

Available on

Pricing from$0.08/ 1M input$0.49/ 1M output$0.04/ 1M cache read524Kcontext

Privacy

ZDR
Training
No
Region
EU

An OBLITERATUS refusal-removed Qwen3.8 derivative with restored visual components.

Available on

Pricing from$0.17/ 1M input$1.19/ 1M output$0.15/ 1M cache read524Kcontext

Privacy

ZDR
Training
No
Region
EU

Heabsy's own refusal-removed Qwen3.8 27B build, with a published 524K context.

Available on

Pricing from$0.67/ 1M input$2.12/ 1M output$0.12/ 1M cache read524Kcontext

Privacy

ZDR
Training
No
Region
EU

A creative-writing and roleplay fine-tune of Qwen3.8 27B released by aifeifei798.

Available on

Pricing from$0.08/ 1M input$0.49/ 1M output$0.04/ 1M cache read524Kcontext

Privacy

ZDR
Training
No
Region
EU

Carina 60B, Multiverse Computing's compact reasoning model built on the gpt-oss architecture, for tool use and agent workflows at very low cost.

Available on

Pricing from$0.02/ 1M input$0.09/ 1M output131Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.30/ 1M input$0.60/ 1M output66Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.22/ 1M input$0/ 1M output16Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.03/ 1M input$0/ 1M output16Kcontext

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing from$0.05/ 1M input$0/ 1M output512context

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing$0.0022 / min

Privacy

ZDR
Training
No
Region
EU

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.

Available on

Pricing$0.0022 / min

Privacy

ZDR
Training
No
Region
EU

Pianissimo, Klang's streaming speech recognition model for Swedish, transcribing live audio into punctuated text as people speak, dialects included.

Available on

Pricing$0.0039 / min

Privacy

ZDR
Training
No
Region
EU

roc, Evroc's enterprise AI agent for writing, data analysis, coding support, and knowledge retrieval over company documents and systems.

Available on

Pricing from$2.76/ 1M input$11.05/ 1M output

Privacy

ZDR
Training
No
Region
EU

Sovereign end to end

The gateway on a European cloud too

The providers on this list are European companies. For the gateway as well, Opper runs as its own instance on evroc, the Swedish cloud, set up with your team.

Sovereign AI on Opper

Frequently asked questions

Which AI models are sovereign?

+
120 models on the Opper gateway run on routes operated by European-owned providers and hosted in Europe as of 2026-11-14, on Lyceum, TensorX, Berget, Munito AI, Mistral, Sference, Multiverse Computing, Inceptron, Melious, GreenPT, evroc, Heabsy, Regolo, Scaleway, Greenference, FlexAI, Infercom, Nextbit, Opper, Reson8, Decision Models, Nodion and Upright. Families include OpenAI's open weights, Mistral, DeepSeek, Qwen, GLM, Kimi, Llama and others. This page lists every one of them with where it runs, what the provider keeps and what it costs.

What makes an AI model sovereign?

+
The route, not the model. A route counts as sovereign here when the company that operates it is headquartered in Europe and inference runs in Europe, so only European companies under European law handle your prompt. Open-weight models from any lab qualify when a European-owned provider runs them, and the lab never receives your data.

Can I run Claude and Gemini on a European-owned provider?

+
Not on Opper today. Claude and Gemini run in the EU on AWS Bedrock and Google Cloud, which keeps processing in the EU with operators headquartered outside Europe. For EU residency, which is what most teams need, they are on the EU list. Every EU-hosted model

How do I make sure calls only reach European-owned providers?

+
Call a route from this list by its model id. On Control Plane, set one Model access rule with provider country and inference location in Europe and it becomes the default for your whole organization: a call to anything else fails with a clear error, and Opper never switches provider silently. Model access rules

Is the gateway European-owned too?

+
On Opper on evroc, yes. Opper runs as its own instance on evroc, the Swedish cloud, so the gateway and the model provider are both European companies. The main Opper gateway runs in AWS Stockholm. Access to Opper on evroc is by onboarding. Sovereign AI on Opper

Read this list from an agent

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/sovereign.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. The lookup 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

Run sovereign AI on Opper

Opper on evroc is set up with your team. Get in touch and we onboard your organization.

Sovereign AI on Opper