Skip to main content
Luminy connects to more than 20 AI providers — from frontier cloud models like Anthropic Claude and OpenAI GPT-4o to fully private local inference via Ollama — all through a single unified interface. You configure each provider once in Settings, then switch between any model mid-session using the model selector in the chat composer. Your chosen model becomes your default for future sessions automatically, and all API keys are stored locally on your device.

​
Configuring a provider

1

Open Settings

Click the gear icon or press ⌘, (macOS) / Ctrl+, (Windows/Linux) to open Settings.
2

Select your provider

Scroll to the provider’s section — for example, Anthropic, OpenAI, or Groq.
3

Enter your API key

Paste your API key into the key field. For OAuth-based providers (GitHub Copilot, ChatGPT), click Sign in instead.
4

Save

Click Save. Luminy verifies the key and the provider’s models become available immediately in the model selector.

​
Selecting a model

Use the model selector dropdown in the chat input area to choose any configured model. The change takes effect immediately — even mid-session. The model you select becomes your default for all new sessions going forward; you can change it at any time.
You can maintain different default models for your main chat session and for Luminy’s background agent (the agent that works on tasks autonomously). Both are configurable in Settings → Model.

​
Provider categories

Local Models

Run Ollama models entirely on your own hardware. No API key, no data sent to the cloud, complete privacy.

Cloud Providers

Access frontier models from Anthropic, OpenAI, Google Gemini, GitHub Copilot, OpenRouter, and 10+ more cloud providers.

Custom Endpoints

Point Luminy at any OpenAI-compatible or Anthropic-compatible API — LM Studio, vLLM, Azure OpenAI, private deployments, and more.

​
Model ID format

Every model in Luminy is identified by a provider-prefix:model-name string. You will see these IDs in the model selector and in any exported session data. 
anthropic:claude-opus-4-5
openai:gpt-4o
gemini:gemini-2.0-flash
groq:llama-3.1-70b-versatile
qwen2.5-coder:7b          ← Ollama models use bare model names (no provider prefix)
openrouter:anthropic/claude-3.5-sonnet
Ollama is the one exception to the prefix rule. Luminy displays local Ollama models by their bare model name — for example, qwen2.5-coder:7b — without any extra provider prefix in the chat UI.

​
All supported providers

ProviderModel ID prefixAPI key requiredNotes
Ollama(bare model name)NoLocal inference on your machine
Anthropicanthropic:YesPrompt caching for lower costs on long sessions
OpenAIopenai:YesAutomatic prefix caching
Google Geminigemini:YesAutomatic prefix caching
GitHub Copilotcopilot:OAuthSign in with your GitHub account
ChatGPTchatgpt:OAuthUses chatgpt.com
OpenRouteropenrouter:YesAccess 100+ models with one key
xAI (Grok)xai:YesGrok model family
Mistralmistral:YesMistral & Mixtral models
Groqgroq:YesHigh-speed LPU inference
DeepInfradeepinfra:YesHosted open-source models
Cerebrascerebras:YesUltra-fast Cerebras inference
Coherecohere:YesCommand R+ and other Cohere models
TogetherAItogetherai:YesBroad open-source model catalog
Perplexityperplexity:YesOnline-search-augmented models
Alibaba DashScopealibaba:YesQwen model family
Venice AIvenice:YesPrivacy-focused inference
Fireworksfireworks:YesFast serverless open-source models
OpenCode Zenopencode-zen:NoFree models, no key needed
Custom (OpenAI-compat)openai-compat:OptionalAny OpenAI-compatible API
Custom (Anthropic-compat)anthropic-compat:OptionalAny Anthropic Messages API-compatible server