NaiveAI Open-Weights Naive-N0.5-Flash, a 309B MoE Model Under MIT
NaiveAI Open-Weights Naive-N0.5-Flash: 309B MoE, Hybrid SWA–DSA 1M Context, MIT
An MIT licence removes the commercial-use restrictions that still accompany many open-weight releases, so teams can fine-tune or ship it in products without negotiating terms. The hybrid SWA–DSA design without any full-attention layers is a notable architectural signal for anyone tracking how long-context efficiency is evolving beyond dense attention. Engineering teams evaluating open-weight alternatives for code generation and research pipelines should benchmark it against comparably sized MoE releases.
NaiveAI has released the weights of Naive-N0.5-Flash under the MIT licence.
The model is a 309B-parameter MoE that activates 15.5B parameters per token.
It uses a hybrid SWA–DSA attention mechanism with no full-attention layers, natively supporting 1M-token context.
The model is built on MiMo-V2.5 and targets coding and AI R&D workloads.
NaiveAI Open-Weights Naive-N0.5-Flash: 309B MoE, Hybrid SWA–DSA 1M Context, MIT
NaiveAI released Naive-N0.5-Flash as MIT open weights: a 309B MoE with 15.5B active, hybrid SWA–DSA native 1M context for coding and AI R&D, built on MiMo-V2.5 without full-attention layers.
This report is categorized under AITOP's "Models" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include NaiveAI, Naive-N0.5-Flash, open-source models, MoE.
Primary source verification: published directly by pandaily with first-party authority.
