Meta 刚刚发布了 Muse Spark 1.2 (xHigh)，在 Agent Arena 上的净提升分数从 0.9% 提升到了 2.1%，Bash Recovery 任务直接翻倍，比前代强了不少。
Meta 发布了 Muse Spark 1.2 (xHigh) 模型。该模型在 Agent Arena 基准测试中实现了 +2.1% 的净提升。相比前代 Muse Spark 1.1，其净提升分数从 0.9% 提升至 2.1%。在 Bash Recovery 任务上，性能从 +5.9% 提升到 +11.4%。
Muse Spark 1.2 (xHigh) by @AIatMeta is now in Agent Arena, with a net improvement of +2.1%! Agent A...
Muse Spark 1.2 (xHigh) by @AIatMeta is now in Agent Arena, with a net improvement of +2.1%! Agent Arena measures models on millions of real-world, long-horizon agentic tasks. We use causal tracing methodology to measure a model's net improvement, indicating how much it improves outcomes relative to the average model. Muse Spark 1.2 improves on Muse Spark 1.1, more than doubling the Net Improvement score from 0.9% to 2.1% (~2.3x). By signal, Muse Spark 1.2 (xHigh) improved significantly in Bash Recovery (CLI error recovery) compared to Muse Spark 1.1, from +5.9% to +11.4%! More detail on performance by signal below. By category, Muse Spark 1.2 (xHigh) is strong at #15 in Chat with +3.7% net improvement. It’s #25 in Code (0.76%) and #26 for Work (0.49%). Congrats again to the @AIatMeta team on this release! AI at Meta @AIatMeta Introducing Muse Code (beta), a terminal coding agent built for long-horizon software engineering, powered by our new Muse Spark 1.2 model. Muse Code plans, implements, and validates complex, multi-file changes across large repositories with persistent sub-agents that solve difficult problems faster, more accurately, and with less intervention. 🧵👇 🔗 View Quoted Tweet 💬 6 🔄 1 ❤️ 34 👀 4954 📊 7 ⚡