Local inference and model providers
Available modalities depend on the model, API and configured account. A bridge does not guarantee access to the provider’s entire catalogue.
Ollama
Local inference and model management.
Reference website
OpenRouter
Access to multiple model providers.
Reference website
OpenAI
API and personal ChatGPT connection, subject to account terms.
Reference website
Anthropic
Access to Claude models through the API.
Reference website
Mammouth AI
Model catalogue through its API.
Reference website
DeepSeek
Inference through the DeepSeek API.
Reference website
Fireworks AI
Models served by Fireworks.
Reference website
Groq
Inference through GroqCloud.
Reference website
Mistral AI
Models available through the Mistral API.
Reference website
Together AI
Access to models hosted by Together.
Reference website
Cerebras
Access to the Cerebras inference service.
Reference website
Google Gemini
Gemini models through their compatible API.
Reference website
xAI
Grok models through the xAI API.
Reference website
NVIDIA NIM
Inference endpoints from the NVIDIA catalogue.
Reference website
Hugging Face
Models through Inference Providers.
Reference website
Cohere
Models through Cohere’s compatible API.
Reference website
Perplexity
Access to Perplexity API models.
Reference website
Voice, music and video
These services complement the multimodal capabilities of model providers.
ElevenLabs
Speech, transcription, music and sounds, depending on the resource.
Reference website
Google Cloud Text-to-Speech
Google Cloud speech synthesis.
Reference website
Azure AI Speech
Speech synthesis through Azure.
Reference website
BytePlus LAS
Video generation through the LAS service.
Reference website
SunoAPI.org
Music and sounds through this third-party intermediary, separate from Suno.
Reference website
Messaging and email
Use your existing channels to communicate with your agents. Each bridge has its own capabilities and requirements.
Nextcloud
Talk for exchanges with agents; file sharing.
Reference website
Matrix
Messaging through a configured Matrix homeserver.
Reference website
Telegram
Exchanges with agents through the Bot API.
Reference website
OneBot 11
Bot protocol using reverse WebSocket and a configured adapter.
Reference website
WhatsApp Business
Messaging through the Cloud API, with a Meta account and configuration.
Reference website
IMAP / SMTP
Read and send email through your mail server.
Reference website
The Matrix logo is a trademark of The Matrix.org Foundation. Its use does not imply endorsement or affiliation.
Execution harnesses
The harness executes tasks; Galaris retains agents, memory and permissions. The internal engine and external harnesses have different capabilities.
Pydantic AI
The framework behind Galaris’s internal harness.
Reference website
Hermes Agent
External harness with tools, memory and skills connected to Galaris.
Reference website
Codex
Managed runtime connected to models and authorised tools in Galaris.
Reference website
Claude Agent SDK
SDK-based harness with the model gateway and Galaris MCP.
Reference website
DeepSeek Harness
DSH runtime adapted to Galaris’s common execution contract.
Reference website
Automation, calendars and metadata
These connections bring processes, events or useful information to agents.
n8n
Business process execution with the n8n engine.
Reference website
iCalendar
Calendar feed reading and related triggers.
Reference website
models.dev
Model metadata; does not provide inference.
Reference website
Protocols to extend your ecosystem
You can also connect services that implement supported protocols. Endpoint and authentication compatibility still need to be checked.
MCP
Tool servers configured in Galaris, with fine-grained agent permissions.
Reference website
Chat Completions
Compatible endpoints for inference and remote harnesses, subject to their capabilities.
Reference website
Choose a connection that fits your work
Configure the required account or server, choose resources and assign permissions in Galaris. Third-party services retain their own terms, pricing and limits. This catalogue covers the project’s active bridges; availability depends on your version and configuration.
Models and harnesses · Tools and channels · APIs and integrations