The Matilda API, on our GPUs
Call Matilda from your own code, served onshore from Melbourne. The SDK reference docs are live, and the dashboard mints your keys and shows your usage by day — sign in with the Matilda account you already have.
Chat, streaming, and files
A self-contained TypeScript client in the OpenAI shape: chat, full-event and durable streaming, structured output, file uploads, and managed auth flows.
- npm install @maincode-ai/matilda-client-sdk
- const client = new Matilda({ baseUrl, accessToken })
- await client.chat.create({ input })
- for await (const event of client.chat.stream({ input }))
- await client.files.upload(file)
Agents, tools, and sessions
Built on the client SDK: named agents, a runner with retry and a client-side tool loop, multi-turn sessions, and durable stream resume.
- npm install @maincode-ai/matilda-agent-sdk
- const result = await run(agent, 'Say hello.')
- for await (const event of stream(agent, prompt))
- await session.run('Show me an example.')
- const toolHandlers: ToolHandlers = { … }
Keys and usage
Sign in with your Matilda account to mint and revoke API keys, and watch tokens and requests by day. Limits are lifted per account while the API is in early access.