Community site maps the Pareto frontier for local LLMs on 2x RTX 3090
EmilPi · reddit · 2026-10-31
The maintainer of locallm.top visualized score-vs-latency Pareto frontiers for local LLMs on 2x RTX 3090 across a community benchmark collection. The frontier hasn't moved since DeepSeek-V4-Flash-Vision-Exp and Qwen3.8-Flash-Next launched; Qwen4-Flash may break it. The site offers domain/language filters and invites contributors.
More from Infra
- Nvidia's enterprise/neocloud segment growing 138.5% could overtake hyperscalers in 2-6 quarters — Beth_Kindig · 2026-10-31
- Nvidia's ACIE segment could overtake hyperscalers in 2-6 quarters, analyst estimates — Beth_Kindig · 2026-10-31
- Uncensored 249GB DeepSeek V4.1 Flash runs on a single Colab GPU at 118 tok/s — Healthy_Essay3388 · 2026-10-31
- Siemens EDA boosts chip design agent score from 55% to 95% with NVIDIA NeMo Gym — NVIDIAAI · 2026-10-31
- EU's AI Energy Has Two Clocks: Grid Flexibility Now, Fusion Later — ingliguori · 2026-10-31
- Matthew Berman open-sources Device Monitor, a free LAN tool to track all your machines — MatthewBerman · 2026-10-31