AI inference providers

Every hosting partner on the Opper gateway, with the per-route privacy and residency detail most directories skip: zero data retention, training posture, GDPR DPA, and where each route is hosted. Filter to the providers that match your requirements.

Opper is the single AI sub-processor for every model on the platform. The model providers behind each route, with their region, retention and training posture, are listed here and, next to the infrastructure services behind the platform, on the Opper trust center.

Looking for the models themselves? Browse the full catalogue → Or see who builds them in the maker directory →

68 providers26 EU-resident39 with zero data retention1127 models →

All hosting providers

Google CloudGoogle's inference platforms — Vertex AI and the AI Studio Gemini API.
12 routes116 models
US, EU, MultiZDR· abuse monitoringZDR on EnterpriseGDPR DPA
AzureMicrosoft Foundry — frontier OpenAI and Anthropic models on a single Azure contract.
12 routes71 models
EU, MultiZDRGDPR DPA
MistralMistral's first-party API — Paris-based, EU-hosted, GDPR-native.
2 routes56 models
EU, MultiZDRGDPR DPA
AWS BedrockAmazon's managed inference platform — the broadest hyperscaler catalogue, EU + US.
9 routes39 models
EU, Multi, USZDRGDPR DPA
Alibaba CloudModel Studio — Qwen lineage hosted in Frankfurt (eu-central-1).
2 routes23 models
EU, MultiZDR· abuse monitoringGDPR DPA
GreenPTDutch platform for energy-conscious inference, open weights served from France and Finland.
1 route19 models
EUZDRGDPR DPA
BergetSwedish AI cloud — EU-only, full GDPR posture by default.
1 route16 models
EUZDRGDPR DPA
TensorXEU-sovereign serverless inference — open-weight models in Dublin & Helsinki with zero data retention by default.
1 route16 models
EUZDRGDPR DPA
evrocEuropean cloud and AI services
1 route14 models
EUZDRGDPR DPA
LyceumBerlin GPU cloud serving open-weight models from European data centres.
1 route12 models
EUZDRGDPR DPA
HeabsySlovak AI platform hosting open-weight models.
1 route11 models
EUZDRGDPR DPA
RegoloItalian green-compute inference — open models served from Seeweb's own datacenters.
1 route11 models
EUZDRGDPR DPA
ScalewayFrench cloud provider serving open-weight models from its own GPUs in Paris.
1 route11 models
EUZDR· abuse monitoringGDPR DPA
GeoddProduction AI inference infrastructure across US and European routes.
2 routes10 models
EU/EEA + USZDRGDPR DPA
GreenferenceFrench inference host for small open-weight models, served from GPU capacity in the EU.
1 route10 models
EUZDRGDPR DPA
Multiverse ComputingSpanish AI company serving compressed and open-weight models through its CompactifAI API.
2 routes10 models
EU + USZDRGDPR DPA
NextbitSpanish AI cloud, open models served from Nextbit's own data center in Spain.
2 routes10 models
EU, MultiZDRGDPR DPA
MeliousGerman platform for open-weight models — EU-resident inference over a network of European hosts.
1 route8 models
EUZDRGDPR DPA
InfercomHigh-speed EU sovereign inference, served from Munich on SambaNova dataflow hardware instead of GPUs.
3 routes7 models
Multi, US, EUZDRGDPR DPA
SferenceManaged inference for open models — EEA-resident, zero data retention by default.
1 route7 models
EUZDRGDPR DPA
InceptronEU-based inference for open-weight models.
1 route6 models
EUZDRGDPR DPA
Munito AIParis-based inference provider serving open-weight models from EU data centers.
1 route5 models
EUZDRGDPR DPA
OpperModels that Opper hosts directly — small specialised models we ship as part of the gateway.
2 routes5 models
EUZDRGDPR DPA
UprightHelsinki company serving its own Eevo decision model.
2 routes2 models
Multi, EUZDRGDPR DPA
NodionStuttgart PaaS provider hosting open-weight models in European datacenters.
1 route1 model
EU
Reson8Amsterdam speech-to-text company running its own GPUs in the Netherlands.
1 route1 model
EUZDRGDPR DPA
DeepInfraLarge open-weight catalogue at scale — US-hosted.
1 route126 models
USZDRGDPR DPA
NovitaBroad open-weight catalogue at competitive prices.
1 route74 models
USZDRGDPR DPA
OpenAIOpenAI's first-party API — direct route to GPT and o-series, US-hosted.
1 route59 models
USZDR· abuse monitoringGDPR DPA
BytePlusByteDance's enterprise AI cloud — first-party route to the Seed and Seedream families.
2 routes33 models
MultiZDR· abuse monitoringGDPR DPA
xAIxAI's first-party API — direct route to the Grok family across EU and US.
6 routes32 models
Multi, USZDRGDPR DPA
Kling AIKuaishou's Kling video, image and audio models, direct from Kling AI in Singapore.
1 route31 models
SingaporeZDR· abuse monitoring
Tencent CloudTencent's model gateway, serving Hunyuan alongside hosted GLM, Kimi, DeepSeek, MiMo and MiniMax models.
3 routes31 models
MultiZDRGDPR DPA
NebiusGPU cloud and managed inference — strong on EU residency for open-weight models.
2 routes28 models
MultiGDPR DPA
falServerless generative media — fast hosting for FLUX, Recraft, and Ideogram image models.
1 route26 models
USGDPR DPA
DigitalOceanServerless inference on DigitalOcean's own GPU fleet
1 route25 models
USZDR· abuse monitoringGDPR DPA
FireworksFast, cost-efficient hosting for open-weight models.
1 route17 models
MultiZDRGDPR DPA
AnthropicAnthropic's first-party API — direct route to Claude, US-hosted.
1 route14 models
USZDR· abuse monitoringGDPR DPA
MorphSpeed-optimised inference for coding agents — US-hosted.
1 route12 models
USZDR· abuse monitoringGDPR DPA
PerplexitySonar — language models with live web search built in.
3 routes12 models
Multi, USZDRGDPR DPA
ElevenLabsProduction-grade voice, speech, and conversational-AI models.
1 route9 models
USGDPR DPA
EmpirioLabsSpecialized AI model hosting for open, proprietary, and custom stacks
1 route9 models
MultiZDRGDPR DPA
GroqLPU-based inference — sub-100ms first-token latencies on open-weight models.
1 route7 models
USZDRGDPR DPA
JarvislabsIndian GPU cloud serving DeepSeek, GLM, Qwen, Gemma and Nemotron from its own GPUs in India.
1 route7 models
IndiaZDR
SCXSovereign AI inference from Australia
2 routes7 models
Australia, MultiZDRGDPR DPA
HeyGenHeyGen's avatar video, lip-sync and voice models, direct from HeyGen's API in the US.
1 route6 models
USGDPR DPA
Baidu QianfanBaidu AI Cloud's model platform, Chinese frontier models served from mainland China.
1 route5 models
Multi
ModalToken-billed inference on Modal-managed shared capacity
1 route5 models
MultiZDRGDPR DPA
PrunaEuropean image and video model lab — compression-optimised inference.
1 route5 models
USZDR· abuse monitoringGDPR DPA
WaferTuned serving for open-weight models — GLM, Kimi and DeepSeek.
1 route5 models
USZDRGDPR DPA
Z.aiZhipu AI's first-party API for the GLM family
1 route5 models
MultiZDR· abuse monitoringGDPR DPA
ArceeArcee's US-hosted API, serving GLM, Kimi and DeepSeek models while its next Trinity generation is prepared.
1 route4 models
USZDR· abuse monitoringGDPR DPA
CloudflareWorkers AI: open-weight models served from Cloudflare's global network.
1 route4 models
MultiZDRGDPR DPA
SoniloUS audio lab: music, sound effects and video scoring from its own models.
1 route4 models
US
MetaMeta's first-party Model API — direct route to Muse Spark.
2 routes3 models
MultiZDR· abuse monitoringGDPR DPA
Moonshot AIMoonshot AI's first-party API — direct route to the Kimi family, globally hosted.
1 route3 models
Multi
Particle AISan Francisco inference provider serving open-weight models from US data centers.
1 route3 models
USZDR
abliteration.aiUS host of abliterated open-weight models for security, red-team and trust and safety work.
1 route2 models
USZDR
CerebrasWafer-scale chip inference — among the fastest tokens-per-second in the industry.
1 route2 models
USZDRGDPR DPA
DeepSeekThe DeepSeek V4 family from its maker's own API
1 route2 models
China
Fish AudioSpeech synthesis and transcription from Fish Audio.
1 route2 models
Multi
Liquid AILiquid AI, Cambridge, Massachusetts: its own d1 decision models on the Liquid API.
1 route2 models
USZDR
MiniMaxMiniMax's first-party API — direct route to the MiniMax M-series.
1 route2 models
MultiZDR· abuse monitoringGDPR DPA
PoolsidePoolside's first-party API — direct route to the Laguna agentic coding models.
1 route2 models
MultiZDR· abuse monitoringGDPR DPA
TypeSafe AISan Francisco AI lab serving Jev, a model that returns calibrated decisions instead of text.
1 route2 models
MultiGDPR DPA
VEEDLondon video platform serving its own talking-avatar and lip-sync models by API.
1 route2 models
MultiGDPR DPA
Bespoke LabsProbabilistic classification, routing and scoring
1 route1 model
Multi
ReveImage model with strong prompt adherence and accurate typography — create, edit, and remix.
0 routes0 models
—

Start building with 700+ models

One API key for every major provider, up and running in minutes.

Get startedView Documentation