Qwen3.6-27B (dense, FP8): quality over speed
Qwen3.6-27B-FP8, a dense 27B model on the DGX Spark (GB10) — every token activates all 27B parameters, so it trades speed for per-token quality.
| Metric | Value |
|---|---|
| Status | ✅ up, no init errors |
| Generation throughput | ~8.3 tok/s steady |
| Prompt throughput | 51–381 tok/s bursts (prefill) |
| KV cache usage | 2–3% (131k context configured) |
| Prefix cache hit rate | 48% → 81%, climbing |
| Active requests | 0–1, all 200 OK |
The slower generation (8.3 vs the 15–25 tok/s of the MoE/smaller models) is expected for a dense 27B — the payoff is noticeably higher quality per token, especially for code and reasoning. Prefix caching climbing to 81% means repeat conversations and code completion feel far snappier than the raw 8 tok/s suggests.
Bottom line: this is my quality dense-reasoning and coding engine, where I’m happy to trade throughput for output I can trust.