PrismML Ternary-Bonsai-27B Q2 Outperforms DeepSeek v4 Flash and Qwen3.6 27B on Select Tests

PrismML's Ternary-Bonsai-27B Q2 model demonstrated unexpectedly strong performance in initial runs, surpassing established models like DeepSeek v4 Flash and Qwen3.6 27B on specific, unnamed tests. The model's capabilities surprised the evaluator, suggesting a notable advancement in its class.