1 October 2026 · Agentic game · arcade
Digger Arcade
Completed a playable arcade Digger game and reported end-to-end verification.
What was checked
- TypeScript check and Vite production build passed.
- 22 of 22 self-tests passed in headless and browser runs.
- 14,400-frame smoke run traversed all five boards without NaN or a stuck player.
- Five levels passed connectivity and floating-bag validation.
- No console errors or warnings were reported in the checked game states.
Limits of this evidence
TTFT and peak RAM were not recorded. Prefill includes cached processing and is not a cold-prefill measurement. Verification results are transcribed from the run report and were not rerun during archival.
Configuration
The model, serving engine and agent used for this execution.
| Field | Value | Detail |
|---|---|---|
| Model · primary | Qwen 3.8 Flash Next · primary | EXL3 2.625 bpw |
| Runtime | Sushi 1.1.0 | inference engine and exact build |
| Harness | Pi coding agent | the agent or client that drove the run |
| Hardware | Apple M5 Pro · 64 GB unified memory | hardware recorded for this execution |
| Workload | Digger Arcade · revision 1 | Interactive coding-agent session; this is not directly comparable with fixed task suites or other challenge prompts. |
| Repetitions | 1 | executions recorded for this workload; one means an indicative result, not an estimate |
Inference measurements
Engine-side rates for this run. Each figure states its own statistic; figures from different runs are not interchangeable.
| Metric | Value | Definition |
|---|---|---|
| Decode · per-request median | 37.2 tok/s | median · n=244 |
| Decode · output-weighted | 34.62 tok/s | output-token-weighted mean · n=244 |
| Prefill · median | 419 tok/s | median; includes cached processing · n=244 |
| MTP acceptance | 69.03 % | aggregate · n=117057 · 80,808 accepted out of 117,057 proposed tokens. |
Agent session counters
Agent-wide usage totals. These include reasoning, tool calls and waiting, and are not model decode speed.
| Metric | Value | Definition |
|---|---|---|
| Session elapsed | 161.7 min | first instruction to final response · n=1 |
| Model requests | 244 requests | total · n=1 |
| Tool calls | 259 calls | total · n=1 |
| Context compactions | 3 compactions | total · n=1 |
Context and token counters
Token totals as reported by the harness or engine. Cached and cumulative figures are not additive with fresh input.
| Metric | Value | Definition |
|---|---|---|
| Maximum prompt context | 113,993 tokens | maximum · n=1 |
Recorded check counters
Counts from the checks recorded with this run.
| Metric | Value | Definition |
|---|---|---|
| Self-tests passed | 22 tests | total · n=1 |
Request records & provenance
244 matched requests in run order, paired by prompt/output token counts. Contains token counts, rates and MTP counters only; raw text, timestamps and internal log identifiers are excluded.
Run ID: 2026-10-01-qwen38-flash-next-sushi-digger-arcade
One execution, with checks transcribed from the retained record. A completed session is not a quality score. Measurement method · All runs