Qwen3.8-Flash-Next: 125B Params, Only 6B Active
Qwen3.8-Flash-Next is an open-weights multimodal mixture-of-experts model from Qwen with 125B total parameters but only 6B active, described as an early preview of the Qwen4 architecture.
What it is
A new open weights multimodal MoE model release from Qwen, positioned as an early architectural preview of the upcoming Qwen4 model family.
What it does
Runs with 125B total parameters but only 6B active per forward pass, which unlike dense models of similar total size gives a significant speed boost while retaining large model capacity.
Why it matters
A 6B active parameter footprint on a 125B total model makes it practical to run large-capacity models locally, and Simon Willison already tested quantized GGUF builds on a DGX Spark via Unsloth's quants.
How to use it
Try quantized versions like the 72.5GB UD-IQ1_S build from Unsloth, or read Qwen's own announcement at qwen.ai.
DeepSeek's New Flash Model Costs $0.14/Mtok and Outranks MiniMax M3
DeepSeek-V4-Flash-0731 is a 304 billion parameter open-weight model with enhanced agentic capabilities, priced at $0.14 per million input tokens and $0.27 per million output tokens.
Inkling: Thinking Machines Lab's first open-weights model
Inkling is Thinking Machines Lab's Apache-2.0 licensed Mixture-of-Experts model with 975B total and 41B active parameters, trained on 45 trillion multimodal tokens.
Kimi K3: 2.8 Trillion Parameters, Weights Coming July 27
Moonshot AI announced Kimi K3, a 2.8 trillion parameter model available now via API, with open weights promised by July 27, 2026.