11. A single long-lived sandbox per project would need provisioning for peak load and months of uptime. ... A single long-lived sandbox per project would need provisioning for peak load and months of uptime. Instead, repositories live on S3-backed storage and each agent runs in its own e 10/13 02:05 ♡
12. G Git provides atomic multi-file commits, authorship, and queryable history. A forge hosts the reposi... Git provides atomic multi-file commits, authorship, and queryable history. A forge hosts the repository, controls who can update it, and adds PR review. Projects keep the first two 10/13 02:05 ♡
13. A Agents collaborating on a long-running project face the same problem as engineers sharing a codebase... Agents collaborating on a long-running project face the same problem as engineers sharing a codebase. Work must outlive ephemeral sandboxes, concurrent writes must not overwrite ea 10/13 02:05 ♡
14. This is how we built the filesystem behind Perplexity Projects as a Git forge for agents. Each agen... This is how we built the filesystem behind Perplexity Projects as a Git forge for agents. Each agent works in its own sandbox and commits to a shared repository where people and ot 10/13 02:05 ♡
15. B Both models are on Hugging Face: https://t.co/JHy1DhFaCT Both models are on Hugging Face: huggingface.co/collections/pe… 💬 0 🔄 0 ❤️ 20 👀 5819 📊 3 ⚡ Powered by xgo.ing 10/08 00:10 ♡
16. On MADQA, 500 questions over 800 PDFs, it reaches 92.4%, the top result among retrievers. On MADQA, 500 questions over 800 PDFs, it reaches 92.4%, the top result among retrievers. 💬 1 🔄 0 ❤️ 15 👀 5941 📊 3 ⚡ Powered by xgo.ing 10/08 00:10 ♡
17. In agentic search, a GPT-OSS-120B agent using the 9B model answers 64.0% of BrowseComp+ questions co... In agentic search, a GPT-OSS-120B agent using the 9B model answers 64.0% of BrowseComp+ questions correctly, 4.9 points ahead of the next ColBERT model, with fewer searches than an 10/08 00:10 ♡
18. On Q2D-Web, about 70,000 production-derived search queries over 190M web documents, the 9B and 0.6B ... On Q2D-Web, about 70,000 production-derived search queries over 190M web documents, the 9B and 0.6B models reach 74.8% and 73.6% Recall @1000 , against 69.3% for nemotron-embed-8b. 10/08 00:10 ♡
19. We're releasing pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images... We're releasing pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images, and pages with a shared embedding space for cross-model querying. Both mod 10/08 00:09 ♡
20. P Perplexity Computer now creates interactive charts and visualizations directly in your thread. For ... Perplexity Computer now creates interactive charts and visualizations directly in your thread. For financial data, Computer uses @tradingview Lightweight Charts for candlesticks, v 10/02 00:12 ♡
