Qwen3.5-9B (FP8-dynamic)

Alibaba / Qwen · Qwen3.5
LLMApache-2.0publicTrusted9.7B params

Ready-made FP8 build (compressed-tensors): 9B dense LLM from the Qwen3.5 generation (02.2026). Strong reasoning, coding and multilingual chat.

Qwen3.5-9B

Dense 9B model of the Qwen3.5 generation (released 02.2026). Outperforms the Qwen2.5 line in reasoning, coding and agents.

Strengths

  • Modern generation: strong reasoning & coding
  • 32K context, multilingual
  • Apache-2.0 — fully commercial

Recommended use

Primary general-purpose assistant model. Runs on a single 24GB GPU at bf16 (weights ~19.3 GB).