Supported Providers
TokenSense supports multiple AI providers. Pick your provider below to see setup instructions and supported models.
Point your workflow at your TokenSense endpoint instead of the provider directly. TokenSense identifies the provider from the model name and routes automatically.
OpenAI
GPT-5, GPT-4.1, o3/o4, DALL-E, gpt-image-1, Whisper, TTS, and embeddings.
AvailableAnthropic
Claude Opus, Sonnet, and Haiku — with native API support.
AvailableGoogle Gemini
Gemini 3.1, 2.5, 2.0, Flash Lite, and Imagen 4 image generation.
AvailableMistral
Mistral Large, Magistral, Codestral, Pixtral, and embeddings.
AvailablexAI (Grok)
Grok 4, Grok Code, and multi-agent models.
Availablefal
Flux 2 Pro, Dev, and Schnell image generation.
Mistral
Add your Mistral API key in Settings → Providers. TokenSense routes automatically when it detects a Mistral model name.
Supported models: Mistral Large, Magistral Medium, Mistral Medium 3, Mistral Small, Codestral, Devstral, Pixtral 12B, Mistral Embed (embeddings).
Setup: Same 3-step process as other providers — add your Mistral API key, then use any Mistral model name in your requests.
xAI (Grok)
Add your xAI API key in Settings → Providers. xAI uses an OpenAI-compatible format natively, so requests pass through with zero translation.
Supported models: Grok 4, Grok 4 Fast, Grok Code Fast, Grok 3, Grok 3 Mini, Grok 3 Fast, Grok 2, and multi-agent models.
Setup: Same 3-step process — add your xAI API key, then use any Grok model name.
fal (Flux image generation)
Add your fal.ai API key in Settings → Providers. fal is supported for image generation only — use the /v1/images/generations endpoint.
Supported models: Flux 2 Pro, Flux 2 Dev, Flux 2 Schnell.
Note: fal does not support chat completions, embeddings, or audio. For chat, use one of the other providers.