SubQ Snap 1.1 (SubQ)
Data: JevBench v1.6.43, published 2026-10-23
Capability 81.9 · API offering, ranked on the API leaderboard
Last measured: 2026-10-22 · measurement release: v1.6.43
Proprietary hosted API (no weights release)
intelligence
75.8
calibration
88.1
speed
84.9
cost
60.7
Composite (secondary)
75.7 · Not ranked on this board
Cost and measurement conditions
$0.020 per 1,000 decisions (estimated)
Vendor's published standard launch tariff, USD 0.04 per million input tokens; output free (read 22 Oct 2026 at https://docs.subq.ai/using-the-api/models/ and https://subq.ai). Counted under the frozen I-1 launch-list-price rule. Capacity-limited public preview, first come first served, with a waitlist once capacity is reached; not openly bookable by everyone. Promotional signup credits are excluded. Own measured usage: 1,939,205 input / 0 output tokens over all 1,500 decisions. Common 1,485-item cost basis: 759,310 input / 0 output tokens, USD 0.020452795 per 1,000 decisions; all-item cost USD 0.051712133 per 1,000. The historical preview row keeps its frozen USD 0.05/M input tariff.
p50 latency: 0.5 s
none (API)
Vendor-hosted api.subq.ai /v1/systemone, served model subq-snap-1.1, measured 22 Oct 2026 from the evaluator host, official per-item latency subset without API adjustment. Full S1200+P300 draw; single same-author second use, provider saw the sealed items in the 13 Oct preview run. Native probabilities, no generated tokens. Hosting implementation and deployed weights were not independently inspected. Inputs retained indefinitely by default; this run used already-retired items.