Ling-3.0-flash offers 124B total parameters with just 5.1B active per token
truecakesnake · reddit · 2026-07-25
Ling-3.0-flash is being offered free in Nous Research’s portal for one week, with the post highlighting its unusual MoE shape: 124B total parameters but only 5.1B active per token.
The thread argues that this low-active configuration is attractive for agent workloads like coding, search, research, and tool use, especially on bandwidth-limited hardware such as Macs, Strix Halo systems, DGX Spark boxes, and large-RAM CPU machines.
Related event: Ant Group Releases Ling-3.0-flash MoE Model(12 posts)→
More from Infra
- MiniMax H3 runs fully on a Galaxy S26: 5-second video with audio in ~7 minutes — BidFunny4862 · 2026-10-27
- Unconventional AI acquires inference chip firm NeuReality in 1,000x efficiency bid — NaveenGRao · 2026-10-27
- Musk: robot RAM demand will dwarf everything else, and 100% of superintelligence compute goes to deep space — elonmusk · 2026-10-27
- RTX 5090 hits $7K as PC gaming community goes into 'denial' over AI boom — PAstynome · 2026-10-27
- Wan2GP lands on Pinokio: 1-click AI video on 6GB GPUs — cocktailpeanut · 2026-10-27
- AWS on agent prompt caching: explicit + implicit patterns to slash token costs — AWS ML Blog · 2026-10-27