>Across a 20-benchmark suite spanning reasoning, math, coding, instruction following, vision, and agentic tool use, Ternary Bonsai 2 27B achieves an aggregate score of 83.9, retaining 98.2% of Qwen3.8 27B’s performance of 85.4. The model reaches up to 143 tokens/second on NVIDIA GeForce RTX 5090.
It's ~6 GB for "98.2% of Qwen3.8 27B's performance". It is significantly faster as well. Remains to be seen how well it works, I had significant issues with the previous Bonsai 27 compared to the parent Qwen model (documented here: https://humanparadox.org/local-vs-frontier-benchmarks-for-my...) but that also graded a bit worse (~95% of Qwen3.6 27B).
>Across a 20-benchmark suite spanning reasoning, math, coding, instruction following, vision, and agentic tool use, Ternary Bonsai 2 27B achieves an aggregate score of 83.9, retaining 98.2% of Qwen3.8 27B’s performance of 85.4. The model reaches up to 143 tokens/second on NVIDIA GeForce RTX 5090.
https://huggingface.co/collections/prism-ml/bonsai-2
It's ~6 GB for "98.2% of Qwen3.8 27B's performance". It is significantly faster as well. Remains to be seen how well it works, I had significant issues with the previous Bonsai 27 compared to the parent Qwen model (documented here: https://humanparadox.org/local-vs-frontier-benchmarks-for-my...) but that also graded a bit worse (~95% of Qwen3.6 27B).