Model monitor
MiniMax M2.7 Reference
minimax/minimax-m2.7 · via openrouter
As of 2026-08-05: reference route · 0 complete panels attest measurement across 1 task families.
Served route, trailing 30 days: median latency 3.8 s · p95 4.6 s across 8 probe calls · served by Mara (8), unrecorded (2). Observed at panel time — a few calls a day, not continuous uptime.
Reference route: this model is the co-probed confound control. It never opens a public incident — its record attests measurement, not verdicts.
Certified claim, per family: H0: E[panel_mean_t | prior completed panels] ≥ frozen_null_floor. Fixed panel composition does not assume or establish temporal independence. Recent panels below link to their hash-committed evidence.
Task family
Code reasoning
- Frozen null floor
- 0.700
- Anytime level α
- 0.050
- Completed panels
- 0
This route is watched. Its record begins at the first complete panel — nothing is backfilled, including for us.
Epoch & design record
- Current epoch
17d9ded1-4381-4d83-9292-4e120bf1d079- Opened
- 2026-08-05 12:27 UTC
- Previous epoch
87f9507e-e714-42c6-9d16-71fb0235e292- Transition
- checker_refresh
- Design hash
bc3668bd52d9d48803b744c2c90f4ed816d7e394525b32b4219f6bab5c9c6df2- Panel hash
827de544c7641d140ac37e321bdb2578a41e291915f519183abbc5bd68153e81- Pool hash
cbfba66601ff51f43c83017d6de9e49988afcee72235f5dca89d705f4f8bb4cc- Panel size
- 5
- Terminal failure
- zero — a failed member scores 0 and stays on the record
- Pool version
code_reasoning-v4- Checker version
code_reasoning-v2- Reference mean μ₀
- 0.700
- Minimum effect δ
- 0.000
- Evidence log e
- 0.00 · max 0.00 · alarm at 3.00 (= ln 1/α)
- Raw scores kept
- 0
The persisted monitor state is the authority a certificate replays byte-for-byte; the chart band above is a faithful replay of the same recorded panel means. Spend sums completed probe runs and includes reserved estimates where a run’s true cost was unrecoverable.