# Meta's Muse Code Claims 82.9% on Terminal-Bench, a Number the Verified Leaderboard Does Not Show

Independent testing by Vals ranks the same model 14th on a common-harness version of the benchmark while placing it fifth overall on cost-adjusted performance at $0.69 per test.

- Published: 2026-08-09T07:02:50.434Z
- Canonical: https://polylog.news/ai/2026-08-09/meta-s-muse-code-claims-82-9-on-terminal-bench-a-number-the
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Meta AI](https://ai.meta.com/blog/introducing-muse-spark-meta-model-api/), [CNBC](https://www.cnbc.com/2026/08/05/meta-debuts-muse-code-to-take-on-anthropic-and-openai-.html), [Kingy AI benchmark review](https://kingy.ai/blog/muse-code-muse-spark-1-2-benchmarks-verified/)

Meta launched Muse Code on August 5, a terminal-based coding agent powered by a coding-tuned model it calls Muse Spark 1.2, entering the market held by Anthropic's Claude Code and OpenAI's Codex. The agent plans edits, runs commands, valida…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-08-09/meta-s-muse-code-claims-82-9-on-terminal-bench-a-number-the (subscription information: https://polylog.news/pricing).