Hy4-Flash local model teased: 224B params with 12B active, one-shot generation in 32 minutes
LiTianleli · x · 2026-11-22
Early tester loktar00 teased Hy4-Flash, a new model that runs locally: a MoE with 224B total parameters and only 12B active. He got early access and ran several tests.
His "bee on a sunflower" prompt completed in a single turn, one shot, in 32 minutes — which he calls "CRAZY for this size of a model." He plans to post more results throughout the day.
More from Models
- Blogger claims Gemini 4 Neon beats GPT-6.2 Astra and Claude Opus 5.5 in tests — lyraxana · 2026-11-22
- Simon Willison: Gemini 4 Hydrogen 'xhigh' thinking costs 6x more but is far better — film_girl · 2026-11-22
- Mathematicians confirm Claude-assisted proof of the Riemann hypothesis is correct — scientificamerican · 2026-11-22
- Researcher argues LLM companies have no moat as Gemini 4 Neon kills the data flywheel thesis — adam_karvonen · 2026-11-22
- Anthropic releases FlashProve with Lean formalization, proving 400B MoE inference on one H100 — jiqizhixin · 2026-11-22
- Gemini users say Low thinking still burns 5-hour quotas on basic prompts, want a non-thinking option back — patg84 · 2026-11-22