577 3 weeks ago

Text-only Qwen3.5 9B GGUF files. Default: q4_k_m. TurinTech models, built with Artemis. Measured Q4_K_M: Mean KLD 0.0145, Same top p 94.0% vs BF16 on WikiText-2 test; protocol rtx5090-2026-10-23.

tools thinking
ollama run turintech/qwen3.5-9b-artemis

Applications

Claude Code
Claude Code ollama launch claude --model turintech/qwen3.5-9b-artemis
OpenCode
OpenCode ollama launch opencode --model turintech/qwen3.5-9b-artemis
Hermes Agent
Hermes Agent ollama launch hermes --model turintech/qwen3.5-9b-artemis
OpenClaw
OpenClaw ollama launch openclaw --model turintech/qwen3.5-9b-artemis

Models

View all →

Readme

Qwen3.5 9B Artemis

Qwen3.5 9B for local chat and tool calls on a GPU with room for a 5.7 GiB peak.

  • 5.661 GB download for the default file, Qwen3.5-9B-Artemis-Q4_K_M.gguf.
  • Its next token matches the original’s on 94.0% of WikiText-2 tokens, against 92.9% for Unsloth’s Q4_K_M (5.681 GB). That text also chose the file.

Ollama

latest selects q4_k_m.

ollama run turintech/qwen3.5-9b-artemis:q4_k_m

This package holds the main file only. For images, use Ollama 0.40.11 or newer and pull from Hugging Face, which brings the repository’s vision projector along:

ollama run hf.co/turintech/Qwen3.5-9B-Artemis-GGUF:Q4_K_M

Results

Our default file against Unsloth’s Q4_K_M, the closest public file in size, and the original. Same next token is on WikiText-2, the text the files were chosen on; the other columns are tasks no selection run used.

File Size Same next token Tool calls (of 258) IFEval (of 541) GSM8K (of 250)
Ours, Q4_K_M 5.661 GB 94.0% 221 456 238
Unsloth Q4_K_M 5.681 GB 92.9% 213 445 241
BF16 original 17.92 GB reference 216 459 242

The files were chosen on closeness over WikiText-2, so that column is not an independent test. On the independent tasks no paired test separated our files from the original or the matched public files. How it was tested.

Files

Every file we publish for this model. Sizes are download sizes, not memory.

File Size Use it for Ollama tag
Qwen3.5-9B-Artemis-Q4_K_M.gguf 5.66 GB Default q4_k_m (latest)
Qwen3.5-9B-Artemis-Q5_K_S.gguf 6.16 GB Larger alternative q5_k_s
Qwen3.5-9B-Artemis-IQ4_XS.gguf 5.17 GB Smaller download iq4_xs
  • The Ollama tag runs our package, which holds the same file: ollama run turintech/qwen3.5-9b-artemis:<tag>.

The vision projector is optional, for image input. There is no MTP file.

Limitations

  • No task gain or equivalence established: no paired test separated it from BF16 or Unsloth Q4_K_M on the independent tasks.
  • Unsloth UD-Q4_K_XL (5.97 GB) has lower KL; in the GB10 replay this file’s WikiText PPL ratio was 1.012493 versus 1.005114 for Unsloth Q4_K_M.
  • The numbers come from llama.cpp, not from Ollama or LM Studio.

Licence

Apache-2.0, as released by Qwen (licence). See LICENSE and NOTICE. This is an independent TurinTech conversion, not affiliated with or endorsed by Alibaba Cloud or the Qwen team.