One allowlist · All three products

Bring a key from any of these.

This is the actual supported-provider allowlist Tempr enforces server-side today — not a wishlist. The same list applies to Chat, the CLI, and the Gateway: add your own API key for any of them and Tempr talks to that provider directly on your key.

See pricing How key storage works

First-party — the model creator's own API · Inference — hosts other companies' open-weight models on its own infrastructure · Cloud — a cloud platform hosting first-party models under its own managed service · Aggregators — one key, routed to many providers through a single API.

OpenAIOpenAIFirst-party
AnthropicAnthropicFirst-party
Google GeminiGoogle GeminiFirst-party
DeepSeekDeepSeekFirst-party
xAIxAIFirst-party
MistralMistralFirst-party
Meta AIMeta AIFirst-party
CohereCohereFirst-party
OpenRouterOpenRouterAggregator
Together AITogether AIInference
Fireworks AIFireworks AIInference
Hugging FaceHugging FaceInference
AI21 LabsAI21 LabsFirst-party
Aleph AlphaAleph AlphaFirst-party
Alibaba CloudAlibaba CloudFirst-party
CodestralCodestralFirst-party
Inception LabsInception LabsFirst-party
Meta (Llama API)Meta (Llama API)First-party
MiniMaxMiniMaxFirst-party
Moonshot AIMoonshot AIFirst-party
MorphMorphFirst-party
PerplexityPerplexityFirst-party
Public AIPublic AIFirst-party
Sarvam AISarvam AIFirst-party
Xiaomi MiMoXiaomi MiMoFirst-party
Z.aiZ.aiFirst-party
AWS BedrockAWS BedrockCloud
Azure OpenAIAzure OpenAICloud
Azure AI FoundryAzure AI FoundryCloud

More providers are added to the allowlist regularly — if yours isn't listed yet, let us know and we'll look at adding it.

One key, three ways to use it

Whatever you bring, it works everywhere — chat in your IDE, the same agent in your terminal, and an OpenAI-compatible API for your own apps.

Explore Chat Explore CLI Explore Gateway