Inferam documentation
Inferam serves an open-weights model to your coding agent. It speaks the three wire formats coding agents already use: OpenAI Chat Completions, OpenAI Responses and Anthropic Messages. Your agent works with it unchanged.
Inferam is in private beta: access is by invite, and it’s free while the beta runs.
Start here
Section titled “Start here”- Quickstart: from invite to your first agent session and your first API call.
- The inferam CLI: sign in once, then run any supported agent on Inferam with
inferam <agent>. - Coding agents: setup for DeepSeek Harness, Claude Code, Codex, opencode and pi, with the CLI or by hand.
- API reference: base URL, authentication, limits, the three formats, usage fields and errors.
- The model: context window, reasoning modes, images and known limitations.
- Privacy and data handling: exactly what we store, and what we never store.
At a glance
Section titled “At a glance”| API base URL | https://api.inferam.ai |
| Formats | /v1/chat/completions, /v1/responses, /v1/messages |
| Model | DeepSeek-V4.1-Flash-1017 (always check /v1/models) |
| Context window | 262,144 tokens |
| Status | status.inferam.ai |
| Dashboard | inferam.ai/keys and inferam.ai/usage |