Add pre-bench Strata snapshot for husky-bench

Passive snapshot taken before the husky-bench run (15.4 h uptime,
1,122 requests), to compare against after the bench.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

benchmarks/live/2026-10-26T2344Z-pre-bench/status.json

@@ -0,0 +1 @@
+{"service": "strata", "model": "qwen3.8-flash-next-iq3_s", "loaded": true, "auto_load": true, "structured_output": {"formats": ["json_object", "json_schema"], "method": "prompt_and_validate", "constrained_decoding": false, "stream_buffered": true}, "engine": "0.1.51", "started": 1793002959, "uptime_s": 55329, "cache_max_tokens": 131072, "context": {"native": 131072, "max_positions": 131072}, "concurrency": {"serving": 1, "requested": 1}, "dialects": ["/v1/chat/completions", "/v1/messages", "/v1/responses"], "vision": {"enabled": true, "available": true, "error": null}, "image_generation": {"enabled": false}, "activity": {"requests": 1122, "in_flight": 0, "last_request_at": 1793058263}, "last_timings": {"cache_n": 0, "prompt_n": 9595, "prompt_ms": 5047.3, "prompt_per_token_ms": 0.526, "prompt_per_second": 1901.0, "predicted_n": 2329, "predicted_ms": 11779.1, "predicted_per_token_ms": 5.058, "predicted_per_second": 197.7, "draft_n": 1938, "draft_n_accepted": 1851, "at": 1793058263}, "vram": {"elastic": false}, "machine": {"at": 1793058288, "gpu": {"name": "NVIDIA GeForce RTX 3090 + NVIDIA GeForce RTX 3090", "used_mib": 48596.0, "total_mib": 49152.0, "util_pct": 0.0, "temp_c": 44, "power_w": 36.631}, "ram": {"used_gib": 81.1, "total_gib": 110.0}}}

benchmarks/live/2026-10-26T2344Z-pre-bench/summary.md

@@ -0,0 +1,64 @@
+# Strata live snapshot, 2026-10-26 23:44 UTC
+
+Model `qwen3.8-flash-next-iq3_s`, engine 0.1.51. Passive counters from normal use, not a benchmark: mixed workloads, so compare like with like.
+
+| Metric | Value |
+|---|---:|
+| Uptime (h) | 15.4 |
+| Requests | 1,122 |
+| Decode tok/s (all output) | 139.3 |
+| Prompt tok/s (all reads) | 1,768.0 |
+| Prompt tok/s (cold, >=4K, recent) | 2,958.8 |
+| Draft acceptance % | 77.5 |
+| Prompt tokens from cache % | 97.0 |
+| Expert cache hit % (recent) | 99.1 |
+| Experts over PCIe % (recent) | 0.1 |
+| Output tokens | 2,143,388 |
+| Prompt tokens read | 2,321,102 |
+
+## Recent requests
+
+| Prompt | Reused | Read | Prompt tok/s | Output | Decode tok/s | Draft acc % | Hit % | Finish |
+|---:|---:|---:|---:|---:|---:|---:|---:|---|
+| 9,595 | 0 | 9,595 | 1,901.0 | 2,329 | 197.7 | 95.5 | 99.1 | stop |
+| 25,890 | 0 | 25,890 | 2,964.4 | 2,390 | 130.6 | 71.7 | 98.6 | stop |
+| 32,631 | 0 | 32,631 | 3,531.3 | 21,158 | 177.5 | 90.9 | 99.0 | stop |
+| 61,184 | 60,110 | 1,074 | 547.5 | 202 | 107.7 | 50.5 | 99.6 | stop |
+| 454 | 0 | 454 | 709.6 | 205 | 156.6 | 83.1 | 98.3 | stop |
+| 753 | 0 | 753 | 699.0 | 2 | 44.2 | 80.0 | 95.4 | stop |
+| 60,115 | 59,999 | 116 | 261.5 | 1,054 | 118.2 | 61.7 | 98.9 | stop |
+| 59,791 | 59,754 | 37 | 190.8 | 206 | 121.9 | 64.2 | 99.6 | stop |
+| 59,511 | 59,377 | 134 | 301.4 | 244 | 129.0 | 67.1 | 99.4 | stop |
+| 59,306 | 59,263 | 43 | 199.4 | 72 | 152.6 | 79.0 | 99.7 | stop |
+| 59,120 | 59,078 | 42 | 183.5 | 143 | 121.8 | 61.5 | 99.7 | stop |
+| 58,545 | 58,508 | 37 | 205.7 | 531 | 116.5 | 60.2 | 98.9 | stop |
+
+## Engine settings
+
+| Setting | Value |
+|---|---|
+| `version` | 0.1.51 |
+| `max_context` | 131072 |
+| `kv` | int8 |
+| `kv_resident` | 32768 |
+| `expert_slots` | 17889 |
+| `expert_slots_primary` | 9508 |
+| `expert_cache_mib` | 34724 |
+| `expert_cache_primary_mib` | 17542 |
+| `arena_mib` | 47962 |
+| `spec` | 6 |
+| `mtp_max` | 4 |
+| `spec_min_p` | 0.50 |
+| `drafting` | mtp |
+| `prefill_cap` | 8192 |
+| `prefill_stages` | 2 |
+| `prefill_pipe` | 0 |
+| `pcie_frac` | 0.34 |
+| `conversation_cache_mib` | 8192 |
+| `vram_free_mib` | 455 |
+
+## Hardware at snapshot
+
+- GPU0: 43°C, 23 W, 23.7/24.0 GiB, util 0%
+- GPU1: 44°C, 13 W, 23.8/24.0 GiB, util 0%
+- Guest RAM: 81.1/110.0 GiB