On Oct 7 LiteLLM open-sourced Moyai (BerriAI/moyai), the self-hosted background coding agent its team runs daily: tasks from Slack or the browser run in isolated cloud workspaces that edit code, run tests and open PRs. Harnesses (Claude Agent SDK, Codex, Hermes, OpenCode, Deep Agents, Tool Loop) are switchable per session and inference routes through LiteLLM's 100+ providers. LiteLLM says replacing Devin cut a 31-day bill from $101,872 to about $21,700.
Key Takeaways
- ✓Cost: $101,872 on Devin vs ~$21,700 on Moyai (~$700/day) over the same 31 days, a claimed 79% / ~$80,172 saving (launch post)
- ✓Pluggable harnesses: Claude Agent SDK by default (prompt caching on), plus Hermes, Codex, OpenCode, Deep Agents and Tool Loop (docs)
- ✓Inference via LiteLLM across 100+ providers; switch GPT-6 Astra / Claude Opus 5.5 / GLM-5.3 mid-session with per-teammate spend attribution
- ✓Provider keys stay server-side and never reach the sandbox; each session gets its own terminal, filesystem and browser
- ✓Caveat: ~58 stars at publish time and no LICENSE file detected by GitHub; confirm terms before commercial use
Key Decision Metrics at a Glance
Not enough VRAM? Compare cloud API and self-hosting costs
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
LiteLLM open-sourced Moyai, the self-hosted background coding agent it runs in production, after its internal Devin bill hit $101,872 in one month. Each task runs in a durable isolated cloud workspace (terminal, filesystem, browser) that edits code, runs tests and prepares a PR; users can correct it mid-task, resume in Slack, or fan large jobs out to parallel workers. The harness is selectable per session: Claude Agent SDK by default with prompt caching, or Hermes, Codex, OpenCode, Deep Agents and Tool Loop. All inference goes through LiteLLM's 100+ providers with per-teammate spend attribution, and provider keys stay server-side away from the sandbox. No task-success benchmarks are published; the only figure is self-reported cost: ~$21,700 (about $700/day) vs $101,872 on Devin over the same 31 days, a claimed 79% saving. Setup: clone BerriAI/moyai, uv sync --frozen, configure .env, and deploy to Modal with uv run python deploy_modal.py against a LiteLLM gateway. Note the repo had ~58 stars and no LICENSE file detected by GitHub at publish time.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.