Meta Ships Its Most Capable Open-Weight Models, with Production Inference Running on SGLang
ying11231 · x · 2026-10-16
SGLang team confirmed Meta released its most capable US open-weight models today, and Meta's production inference runs on SGLang — "the best open model on the best open-source inference engine."
The quoted post details Muse Spark Open, built to plan, act, and use the computer across agentic tasks, letting users build a personal AI agent on a fully open stack: NVFP4 MegaMoE and sequence parallelism for responsiveness, the DFlash2 draft model for faster agentic/coding work, and fine-tuning via Miles. Partners include Meta, NVIDIA, AMD, Intel, and RadixArk.
Related event: Meta Open-Sources Muse Spark Open: 816B Parameters, 1M Context(81 posts)→
More from coding & agent
- Four Gemini 4 Argon agents beat Minecraft's Ender Dragon autonomously in 27 hours — Maxence Faldor · 2026-10-16
- Tencent's Terminal-RSI builds recursively self-improving terminal agents, hitting 19.84% on Terminal-Bench 4.0 — tencent · 2026-10-16
- Tsinghua's AutoSFT uses multi-agent collaboration to automate SFT data research — TsinghuaNLP · 2026-10-16
- ByteDance's Athanor: autonomous agents post-train a small model to match its official tuned version — Yufa Zhou · 2026-10-16
- NVIDIA's EdgeWise cuts long-horizon agent inference costs up to 62% via selective cloud delegation — nvidia · 2026-10-16
- Intern-Decision: 0.8B-4B multimodal decision models output structured choices in one forward pass — internlm · 2026-10-16