RSI-Jev v6.1-VL 27B
Data: JevBench v1.6.39, published 2026-10-22
Capability 73.5 · Not ranked in the current release
cost 6.1× Jev (v1.5 reference)
Last measured: 2026-10-10 · measurement release: v1.6.5
Apache-2.0 (model card); source Shanghua-Gao/RSI-Jev@fc982bbd MIT
intelligence
61.6
calibration
85.3
speed
84.7
cost
31.1
Composite (secondary)
21.6 · Composite rank #86
Cost and measurement conditions
$0.197 per 1,000 decisions (estimated)
ESTIMATE (*est.), base-model market reference. The declared base Qwen/Qwen3.8-27B resolves in the frozen 25 Sep 2026 snapshot through pricing_v15 BASE_REFERENCES to qwen/qwen3.8-27b at USD 0.42 per 1M input and USD 3.00 per 1M output (verified before the pod was rented; r75 COST-BASIS.json). This row reports a real output token count - 1 token per readout, 1,471 in total - so the Gutsy zero-output rule does NOT apply and the output price is charged as measured. This row's own usage on draw v1.6-regular-20261010-a2: 694,381 input and 1,471 output tokens over 1,500 decisions.
Measured p50 latency: 0.123 s (on-GPU, no network, loopback; not directly comparable with API latency, which we time end to end from our measuring host, network included)
Ranking and cap use: 0.40 s (est. incl. network (2x + 150 ms))
Self-hosted model: measured p50 0.123 s on-GPU, no network (timed on our GPU over loopback). Not directly comparable with API latency, which we time end to end from our measuring host, network included. The ranking and latency cap use the same estimate for every self-hosted row: 2 × measured + 150 ms = 0.40 s. Model authors can ask us to time a hosted endpoint instead (on request).
x2 + 0.15 s (assumption, not measured)
evaluator-owned Lium H100 80GB pod (36 min, USD 1.95/h), sealed per chief-of-staff decision #16415 on Florian's YES: the author's own serve.cli inside a loopback-only network namespace (unshare -n; DNS and outbound TCP both proven dead from inside the run namespace), started with env -i and OPENAI_API_KEY empty, plus a pod-level iptables egress cut installed and verified before the sealed input was uploaded. HF_HUB_OFFLINE/TRANSFORMERS_OFFLINE set only to block the download fallbacks the r75 review found; no source line was edited. Server pointed at the LOCAL PINNED SNAPSHOT DIRECTORY, not a repo id, so no revision could resolve to main. Unchanged defaults, --effort auto, 32,768-token cap, no truncation. 54,047 MiB of 81,559 MiB used.