All tasks / Tools
Muse Code
Meta's terminal and CI coding agent built for Muse models, with a headless mode, session protocol and TypeScript SDK.
- Meta
- Cloud
- From $5/mo
Where it ranks
- SCodingCoding agents · Cloud · 68 on Artificial Analysis Coding Agent Index (v1.5)
Ranked by its own benchmark result where it has one; otherwise by the best model it offers.
Facts
- Vendor
- Meta
- Approach
- Coding agents
- Where it runs
- Cloud only: needs the vendor’s service
- Interface
- Command line, API
- Skill level
- Intermediate: some setup or configuration
- Price
- From $5/mo
- Pricing notes
- Monthly plans Everyday / High / Power / Maximum Usage (Power adds Muse Wonder Full); Meta's page shows no prices; 'subscriptions starting at $5' per RuntimeWire (2026-08-31, out of beta); also pay-as-you-go per token. No free tier found.
- Source code
- Closed source
- Last verified
- 19 Oct 2026
Price from runtimewire.com, 19 Oct 2026.
Models it gives you
- BMuse Spark 1.3Meta · closed
- BMuse WonderMeta · closed
Its own benchmark results
| Benchmark | Result | Entry on the leaderboard | As of | Source |
|---|---|---|---|---|
| Artificial Analysis Coding Agent Index (v1.5) | 68 | Muse Code + Muse Wonder Full (max)AA label 'Muse Code - Muse Wonder Full (max)'; Coding Agent Index v1.5; value = indexScore x 100 (0.6803); behind Claude Code + Sonnet 5.5 (max) 68.4 and Devin Fusion CLI 68.1; read from the page's embedded data; no last-updated date shown | 19 Oct 2026 | artificialanalysis.ai |
| Terminal-Bench 4.0 | 14.6% | Muse Code + Muse Spark 1.3reasoning effort: xhigh; rank 32 of 35 entries; raw accuracy 14.55 ± 2.94 (95% CI half-width); entry date 2026-09-02; agent org: Meta; read via the site's leaderboard-read API (page is client-rendered) | 19 Oct 2026 | tbench.ai |