A LoRA fine-tuned version of
Qwen3.5-9B (dense), purpose-built for tool calling, error recovery, and system prompt following. Runs on
8GB VRAM.
Part of the Wrench family — the 35B sibling scores 82% on the
Berkeley Function Calling Leaderboard (BFCL).
1ollama create wrench-9b -f Modelfile
2ollama run wrench-9b
1./llama-server -m wrench-9B-Q4_K_M.gguf --jinja -ngl 100 -fa on \
2 --temp 0.4 --top-k 20 --top-p 0.95 --min-p 0 --presence-penalty 1.5 -c 8192
1npm install -g @clanklabs/clank
2clank setup
3# Set primary model to "ollama/wrench-9b" in config
All training data is published and auditable:
ClankLabs/wrench-training-data1,356 examples (1,251 base + 105 frontier-targeted) across 15 categories.
