Humansand mid/post-trained Persimmon on thousands of Blackwell GPUs from NVIDIA's 550b Nemotron 3 Ultra
gharik · x · 2026-09-11
- Humansand revealed that its new Persimmon model was mid- and post-trained starting from NVIDIA's 550b Nemotron 3 Ultra, using thousands of Blackwell-generation GPUs on a recently built cluster with partners including NVIDIA and DeepInfra.
- Persimmon is an early release; tool use, longer horizons, and more capabilities are still on the roadmap.
More from Infra
- 1:26 continuous aerial AI video made entirely on a Mac with MiniMax H3 — cocktailpeanut · 2026-09-11
- KV cache gets QAT too: why this model beats others at fp4 KV cache — stochasticchasm · 2026-09-11
- Commentary: Anthropic loads shift to Google plus AWS slice, OpenAI doubles down on Azure — ericwdolan · 2026-09-11
- Nvidia claims Vera Rubin delivers 50X throughput per MW and 35X lower token cost vs Blackwell Ultra — Beth_Kindig · 2026-09-11
- Epoch AI: GPT long-context latency scales quadratically, matching price jumps — Jsevillamol · 2026-09-11
- RTX 3090 mini-bench: ninfer cuts TTFT from 3.4s to 29ms, prompt processing ~76x faster — milkipedia · 2026-09-11