Qwen 3.8 Flash Next
A lightweight, efficient inference model in the Qwen series that incorporates an Engram mechanism as an architectural improvement. It aims to enhance intelligence while maintaining a small model size, making it suitable for local deployment scenarios.
Core Facts
Timeline (last 90 days)
The open-source project StraightA claims to be able to run the 125B-parameter Qwen 3.8 Flash Next model on a consumer-grade PC with more than 12G of VRAM, provided the system is equipped with 35 to 55G of RAM.
Strata is an open-source inference runtime capable of running the 125-billion-parameter Qwen3.8 Flash Next model on an RTX 5070 (12 GB VRAM) at an output speed of approximately 90 tokens per second.
A Reddit user ran Qwen 3.8 Flash Next on a hardware configuration featuring dual 3090s and expanded memory.
Qwen 3.8 Flash Next powered by Codex CLI outperformed GPT 5.6 Luna on an actual project.
On the Qwen 3.8 Flash Next model, decoding performance doubled to 52 tok/s.
Qwen 3.8 Flash Next is comparable to DeepSeek R1 Pro on certain tasks.
Qwen 3.8 Flash Next adopts a Mixture of Experts (MoE) architecture.
Qwen 3.8 Flash Next is permanently free to download, deploy, and use commercially, with no subscription fees required.
Qwen 3.8 Flash Next is completely free and open-source.
On a dual AGX Orin system, Qwen 3.8 Flash Next achieves an inference speed of 38 tokens/second.
2 more timeline events
All Facts (9)
Qwen 3.8 Flash Next adopts a Mixture of Experts (MoE) architecture.
80%UnverifiedQwen 3.8 Flash Next is permanently free to download, deploy, and use commercially, with no subscription fees required.
60%UnverifiedOn a dual AGX Orin system, Qwen 3.8 Flash Next achieves an inference speed of 38 tokens/second.
60%UnverifiedQwen 3.8 Flash Next implements QSA (Qwen Sparse Attention), reducing the overhead of long-context processing through block-level sparse attention.
60%UnverifiedThe open-source project StraightA claims to be able to run the 125B-parameter Qwen 3.8 Flash Next model on a consumer-grade PC with more than 12G of VRAM, provided the system is equipped with 35 to 55G of RAM.
50%UnverifiedStrata is an open-source inference runtime capable of running the 125-billion-parameter Qwen3.8 Flash Next model on an RTX 5070 (12 GB VRAM) at an output speed of approximately 90 tokens per second.
50%UnverifiedOn the Qwen 3.8 Flash Next model, decoding performance doubled to 52 tok/s.
50%UnverifiedQwen 3.8 Flash Next adopts a four-branch gated residual mechanism, allowing different types of information to flow through independent channels.
50%UnverifiedQwen 3.8 Flash Next is completely free and open-source.
50%