Not work for agent?

[ui-zeyu]

I use the following command to run llama.cpp:

llama-server --batch-size 4096 --ubatch-size 4096 \
      --model ~/Downloads/Models/Qwen3.6-35B-A3B-BF16-00001-of-00002.gguf \
      --mmproj ~/Downloads/Models/Qwen3.6-35B-A3B-mmproj-BF16.gguf \
      --alias qwen3.6:35b \
      --chat-template-kwargs '{"preserve_thinking": true}' \
      --kv-unified --cache-type-k q8_0 --cache-type-v q8_0 \
      --flash-attn on --fit on \
      --spec-default \
      --temp 0.60 --top-p 0.95 --top-k 20 --min-p 0.00 --presence-penalty 0.00 --repeat-penalty 1.00 \
      --n-gpu-layers all \
      --offline --grammar-file structured_cot_code.gbnf --reasoning-format none

And in codex, it seems not work:

[andthattoo]

Currently; it doesnt work with multi-turn agents. We've created a post in llama-cpp discussions; you can follow the progress there:

ggml-org/llama.cpp#22408