benchmarked Qwable-5-27b-coder on my rtx 5090.
a real-trace coding SFT of Qwen3.6-27b, trained on Claude Fable-5 traces then Kimi 2.7 Coder traces.
every cheap eval agrees it's fine:
~ quality q_avg 93.68 vs base 94.05, flat
~ agentic score 98.61 vs base 98.61, identical to https://t.co/10jZCH7GFC
a real-trace coding SFT of Qwen3.6-27b, trained on Claude Fable-5 traces then Kimi 2.7 Coder traces.
every cheap eval agrees it's fine:
~ quality q_avg 93.68 vs base 94.05, flat
~ agentic score 98.61 vs base 98.61, identical to https://t.co/10jZCH7GFC