NVIDIA ships Nemotron 3.5 Embed, Rerank, and Parse open retrieval models for agents
baseten · x · 2026-10-21
NVIDIA expanded its Nemotron Retriever collection with open, commercially available retrieval models now live on Baseten and Hugging Face: Nemotron 3.5 Embed 8B BF16, 1B BF16, and 1B NVFP4, Rerank 8B BF16, and a 631M-parameter Parse document-extraction model.
- Multimodal retrieval: the models let agents pull evidence across text and visual enterprise data (charts, tables, reports)
- Cost argument: poor retrieval makes agents repeat searches and burn inference tokens; reranking improves evidence quality entering the context
- Target use cases include multi-source reasoning tasks like stockout-risk prediction
Related event: NVIDIA Releases Nemotron 3.5 Retriever Family on Baseten(2 posts)→
More from coding & agent
- Training an original-character LoRA for MiniMax H3: full workflow and pitfalls — DeliciousGorilla · 2026-10-21
- Dev builds open-source Chrome extension unifying Gemini and Antigravity on Windows — iamfivestars · 2026-10-21
- MCP server SDK CVE-2026-111770: shared McpServer leaks responses across clients — Future_AGI · 2026-10-21
- When a Customer's AI Assistant Calls Support, Who Checks What It's Authorized to Do? — IrfanZahoor_950 · 2026-10-21
- Two weeks in SF: fundraising goes binary, top devs now review agents not code — ai · 2026-10-21
- Hugging Face livestream tomorrow: scaling RL and evals to train frontier agents — ben_burtenshaw · 2026-10-21