Alex Getmanearly AI signals
EN

Qwen3.8 Flash Next

Alibaba

Released Updated
Context window
262K
262,144 tokens
Input / Output
image, text, video → text

Timeline

  1. · Open weights · evidence link
  2. · API availability · evidence link
  3. · Weights version · evidence link
  1. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  2. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  3. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  4. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  5. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  6. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  7. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  8. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  9. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  10. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  11. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  12. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  13. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  14. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  15. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  16. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  17. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  18. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  19. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  20. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  21. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  22. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  23. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  24. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  25. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  26. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  27. · Evidence links · GitHub search · Alibaba · pulls · vllm-project/vllm-torchtpu · @DRZDONG
    Excerpt
    Part of the Qwen3.8-Flash-Next model directory (Experimental tier),
  28. · Evidence links · GitHub search · Alibaba · pulls · microsoft/onnxruntime-genai · @tianleiwu
    Excerpt
    context-tier budget adaptation and hardware-aware `runtime_profiles` improvements for Qwen3.8-Flash-Next RC5 (`QMoE`) and GGUF (`MoEPacked`) hybrid offload across 16 GB, 24 GB, 32 GB, and H200
  29. · Evidence links · GitHub search · Alibaba · pulls · TnzGit/dsh-superagent-prototype · @TnzGit
    Excerpt
    - Sanitized local-Agent verification (2026-11-08): Qwen3.8-Flash-Next on a local vLLM dev build had working tool calls and streaming; 8K/32K/64K were exercised without OOM. Short-response decode
  30. · Evidence links · GitHub search · Alibaba · pulls · Azure/azureml-assets · @hanbitmyths
    Excerpt
    | Qwen3.8-Flash-Next | 256K tokens | 61 GiB |
  31. · Confirmed use · Qwen · official article bodies · Qwen3.8-Flash-Next: A New Architecture, Towards Ultimate Cost-Efficiency

Alex's take

Specifications

Qwen3.8 Flash Next is a model from Alibaba with open weights, released on 3 Aug 2026. The context window is 262,144 tokens.

Context window
262,144
Input
image · text · video
Output
text
Weights
Open

Questions

What is the context window of Qwen3.8 Flash Next?

The context window is 262,144 tokens.

Are the weights of Qwen3.8 Flash Next publicly available?

Yes. Alibaba publishes the weights.

When was Qwen3.8 Flash Next released?

On 3 Aug 2026.