Model Detail
Qwen 3.6 27B 4bit
The baseline quant is live but underperformed badly in this first pass, so it is the clearest “not ready” of the trio.
Benchmark score
20/100
Average latency
17.53s
Role
Baseline 27B quant
Strengths
- Live on oMLX/OpenClaw
- Clean endpoint invocation
- Useful as baseline reference
Weaknesses
- Worst score of the trio
- Only 20/100 on quick pack
- Not fit for default routing
Operator read
The baseline quant is live but underperformed badly in this first pass, so it is the clearest “not ready” of the trio.
Source artifacts
Raw machine-readable files for anyone who wants to dig deeper or run their own analysis.