oenogarum (plantwaterer; commercially restricted)v1.7 · 10 s rule
Data: JevBench v1.8.2, published 2026-10-30 · method v1.7 (10 s rule), measured under the rule
Capability 83.9 · Not ranked in the current release
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: 2026-10-30 · measurement release: v1.8.2
Apache-2.0 declared on the model card; no LICENSE file in the repository
intelligence
74.7
calibration
93.1
speed
84.9
cost
55.9
Composite (secondary)
74.4 · Not ranked on this board
Cost and measurement conditions
$0.030 per 1,000 decisions (estimated)
GPU-time estimate from v1.8.0: one monthly-rented RTX PRO 6000 96 GB, EUR 999/month / 730 hours, ECB USD 1.1133/EUR dated 28 Oct 2026, USD 1.5235/hour; average load 70%. Architecture: dense gemma, 31B active parameters; throughput 10106.516129 tokens/s (architecture-class calibration, 29 Oct 2026, efficient-engine prefill; capped at 100,000 tokens/s). The row's own scored cost basis: 735,102 input + 0 output tokens over 1,489 decisions. Generated tokens are counted at the same throughput as prefill. GPU-time cost USD 0.0295320334008779/1,000; old list-price estimate USD 0.0444319543317663/1,000 (google/gemma-4-31b-it at USD 0.09/M input and USD 0/M output). Applied min(old list-price estimate, GPU-time cost) = USD 0.0295320334008779/1,000. Estimated, not a bill. USD 0.0295320334 per 1,000 decisions on the frozen 1,489-item common set of this draw (USD 0.0295061687 over all 1,500 items).
p50 latency: 0.4 s
Self-hosted latency estimate under the shared method (assumption, not measured)
Measured 30 Oct 2026 on an evaluator-owned Verda VM in FIN-03 (Finland, EU), one NVIDIA RTX PRO 6000 Blackwell Server Edition, 97,887 MiB reported, driver 580.178.04, image 24.04.cuda13.0. A first-party loopback HTTP shim invoked the unchanged author library at revision 9b684d7da38eb200fe4c5d11c36e3ce819ea2cfb with verified local artifacts, read-only mounts, empty credentials, offline flags and network cut before sealed input. Python 3.12.12, torch 2.14.1, CUDA 13.0, transformers 5.12.1 and tokenizers 0.22.2 were pinned. Source, runtime and tokenizer hashes were retained. Public smoke used separate warm-ups and fresh timed bodies at 256, 2,048 and 4,096 tokens. Noul, choice, 255-choice, Unicode JSON, special-token and 939/940 structural-boundary probes passed with native/shim probabilities equal. The dynamic 255-choice path was bound-checked, without claiming an exact offline finalist prediction. Unsupported score and over-limit probes returned HTTP 422. These are synthetic observations, not scored latency percentiles. The complete 1,500-request run exited successfully; raw output sha256 d91213504874e50dfafca7c72590bfc1df98c8d60ab6efdad857a896ca6dbc56 was independently verified. Health was ready after the run. The VM was closed and temporary node and local files were shredded. GPU rental spend is separate from the benchmark's model-cost estimate.
Measured latency over 1,500 decisions: p50 0.143 s · p95 0.294 s
0 decisions over 10 s.
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.