Perplexity 发布了新的研究报告，详细介绍了其“Portable Computer”代理，该代理专为本地优先、私密且经济高效的工作而设计。该代理使用经过后期训练的 PPLX 27B 模型，在知识工作和网络研究任务上实现了高准确性。该系统强调对敏感文档进行设备上处理，并在必要时可以升级到前沿顾问，与其它模型和工具相比，展示了改进的性能和成本效益。 AI
影响 通过本地处理和专业模型训练，增强了 AI 任务的隐私性和成本效益。
排序理由 该集群描述了 Perplexity 的新研究和产品发布，详细介绍了一个具有基准性能的新代理和模型。
AI 生成摘要 · Google Gemini · 来自 7 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
Advances in models, chips, or devices will continually improve local-first capabilities. We will open our benchmarks soon. Read the full post: https://t.co/SHYlWB56zA
We post-trained the model to purpose. PPLX 27B is trained inside the Computer harness on synthetic tasks developed from how people actually use Computer, no real user data. https://t.co/0Z16bihCdg
The local model can escalate to frontier advisor. It's user-gated, PII-flagged, text guidance only. On Terminal Bench 2.1, escalation lifts the score from 59.6% to 73.0% at $0.415 per rollout, recovering about three-fifths of the frontier gap at two-thirds of the cost. https://t…
Parsing runs entirely on-device, so sensitive documents never leave the machine. On ParseBench-100, Computer scores 65.1% vs 34.6% for Hermes and 13.9% for Pi, in least time with fewest tokens. https://t.co/GQ590BYEn5
Web research: on 1,266 BrowseComp tasks, Computer reaches 66.7% accuracy vs 50.2% for Pi and 43.9% for Hermes on their respective search providers. Portable uses the least wall time and the fewest tokens. Inference and private documents stay local; only search touches the web. h…
The model and harness are designed together because small models fail in harnesses built for frontier models. Portable is a minimal system prompt, skills that load on demand, connectors as compact CLI tools instead of MCP servers, self-verification, and an always-on sandbox. htt…
New research: Portable Computer is a local-first agent for private and cost-effective work. With an on-device 27B model, our harness scores 82.6% on real knowledge work, beating open-source harnesses Pi and Hermes. Our post-trained PPLX 27B reaches 85.4%. https://t.co/Rb6d47clCI