Artificial Analysis 发布新基准 AA-Briefcase，测试模型在数千个输入文件上的真实任务（电子表格、演示文稿、UI 原型）。Meta 的 Muse Spark 1.1 取得 863 分，与 Gemini 3.5 Flash 和 NVIDIA Nemotron 3 Ultra 持平，较前代提升 232 分。其 rubric 通过率 34.5%，超过 GPT-5.5 (xhigh)，略低于 GLM 5.2 (max)。但演示质量是短板：Presentation Elo 仅 432，较前代下降 67 分，低于 Mistral Medium 3.5。
Muse Spark 1.1 scores 863 on our agentic knowledge work benchmark AA-Briefcase, matching Gemini 3.5 Flash overall, but records a comparatively weaker Presentation Elo of 432, below the previous Muse Spark model
Our new agentic knowledge work benchmark, AA-Briefcase, tests models on realistic tasks across thousands of input files, requiring deliverables such as spreadsheets, presentations, and UI mock-ups. Model performance is measured across three dimensions: binary rubric checks for ground-truth correctness, pairwise grading on analytical quality, and pairwise grading on presentation quality. The AA-Briefcase Elo is a single metric that combines results across all three dimensions
The latest model release from @Meta, Muse Spark 1.1, achieves an AA-Briefcase Elo of 863, placing it alongside Gemini 3.5 Flash and NVIDIA Nemotron 3 Ultra. Its 232-point improvement over the previous Muse Spark model is driven by stronger objective task completion and analytical quality. Muse Spark 1.1 achieves a rubric pass rate of 34.5%, ahead of GPT-5.5 (xhigh) and just behind GLM 5.2 (max)
Presentation quality however remains a relative weakness. Muse Spark 1.1 scores a Presentation Elo of 432, a 67 point decrease from the previous Muse Spark model and slightly below Mistral Medium 3.5
来源：Artificial Analysis · x.com