decide-1.1 (decide.sh, API)v1.7 · 10 s rule
Data: JevBench v1.8.1, published 2026-10-30 · method v1.7 (10 s rule), measured under the rule
Capability 84.0 · API offering, listed but not ranked on the API leaderboard
Measured on the 27 Oct 2026 draw, which is not yet equated to the reference scale; stronger anchor runs for that draw are being measured. Shown unranked until then.
Last measured: 2026-10-29 · measurement release: v1.8.1
Proprietary hosted API (no weights published)
intelligence
75.7
calibration
92.3
speed
90.5
cost
70.2
Composite (secondary)
81.1 · Not ranked on this board
Cost and measurement conditions
$0.0098 per 1,000 decisions (not published)
API launch list price (I-1): https://decide.sh/pricing, applied since 2026-10-29. USD 0.03 per million input tokens; output free (input-only billing), no base model declared and no override. Frozen pricing module validates the operator standard launch list price. Native server usage; 19,926 reported output tokens are unbilled. USD 0.0098373002 per 1,000 decisions on the frozen 1,489-item common set of this draw (USD 0.0177799 over all 1,500 items).
p50 latency: 0.1 s
none (API)
Vendor-hosted https://api.decide.sh/v1/systemone (AWS, CloudFront edge HEL51), model decide-1.1, measured 29 Oct 2026 23:46-23:58Z from the measurement client with the customer's own key, network included (p50 0.094 s, p95 0.88 s wall; 2 of 1,500 over 10 s). Full S1200+P300 set of v1.6-regular-20261027 (input sha256 9b2f32ed...) through the unchanged measurement harness and API adapter (eceea17e / 780f8794); full-set exposure of 1,200 sealed items to provider decide (proxy aws) logged before the first request (pre-registration review). Native System One answer fields on all three question types; served model decide-1.1 on every row; 1,500/1,500 HTTP 200.
Measured latency over 1,500 decisions: p50 0.094 s · p95 0.882 s
2 decisions over 10 s (2 of them answered without error, counted as wrong).
A decision that takes longer than 10 seconds (measured time) counts as a wrong answer and as zero calibration for that item: each type's calibration is multiplied by the share of its decisions answered within 10 s. The limit uses the measured time (raw). The displayed adjusted time for self-hosted rows (2 × + 0.15 s) is an assumption and is not used for the limit.