Bonsai 2 QAT quantized model scores ~91.5% in independent benchmarks

ByteShape independently benchmarked Prism-LM's Bonsai 2 QAT quantized model on Qwen3.8, achieving roughly 91-91.5% on composite benchmarks using Prism's official runtime, indicating strong performance for a quantized model.

2026-09-19 ~ 2026-09-19 · 2 related posts

• Bonsai 2 QAT models score ~91.5% in independent Qwen3.8 quantization comparison — ali_byteshape · 2026-09-19

1 near-duplicate retellings: airesearch12