Ornith-Agents-A1-GGUF

GGUF conversion of tepirale/Ornith-Agents-A1-3.6-35B-A3B-dare_ties for local inference with llama.cpp-compatible runtimes.

About

This model is based on:

• Qwen/Qwen3.5-35B-A3B

Upstream merge sources:

• InternScience/Agents-A1
• deepreinforce-ai/Ornith-1.0-35B

Upstream merge method:

• DARE-TIES

This repository provides GGUF files for practical local use.

Important notes

• These GGUF files are based on an experimental community merge.
• The current release is a text-focused, no-MTP build for compatibility.
• If an MTP version is added later, it should be released as a separate variant.

Files

Model files are named like:

• Ornith-Agents-A1-Q3_K_M.gguf
• Ornith-Agents-A1-Q4_K_M.gguf
• Ornith-Agents-A1-Q5_K_M.gguf

Recommended starting point

If you just want to test the model, start with:

• Ornith-Agents-A1-Q4_K_M.gguf

Example llama.cpp usage

CLI

./llama-cli \
  -m Ornith-Agents-A1-Q4_K_M.gguf \
  --jinja \
  -c 8192 \
  --temp 0.6 \
  --top-p 0.95 \
  --top-k 20 \
  --min-p 0.0

Server

./llama-server \
  -m Ornith-Agents-A1-Q4_K_M.gguf \
  --jinja \
  -c 8192 \
  --host 127.0.0.1 \
  --port 8080

Credits

• Upstream merge: tepirale/Ornith-Agents-A1-3.6-35B-A3B-dare_ties
• Qwen/Qwen3.5-35B-A3B
• Merge sources:
• InternScience/Agents-A1
• deepreinforce-ai/Ornith-1.0-35B
• GGUF runtime ecosystem: llama.cpp

Disclaimer

This is a community GGUF conversion of an experimental merge. Please evaluate it yourself for quality, stability, and suitability for your use case.