Xor 26B-A4B (Juspay, Gemma-4-26B-A4B, bf16)v1.7 · 10 s rule

Data: JevBench v1.8.18, published 2026-11-20 · method v1.7 (10 s rule), re-scored from the recorded per-decision times

Capability 71.2 · rank #75 of 251 on the open-weights board

Last measured: 2026-10-06 · measurement release: v1.6.1

apache-2.0

Base model: google/gemma-4-26B-A4B-itsource

intelligence

57.5

calibration

84.9

speed

90.8

cost

72.1

Composite (secondary)

74.0 · Composite rank #32

Cost and measurement conditions

$0.0085 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: moe gemma, 4B active parameters; throughput 34662.5 tokens/s (architecture-class calibration, 29 Oct 2026, efficient-engine prefill; capped at 100,000 tokens/s). The row's own scored cost basis: 717,195 input + 1,477 output tokens over 1,477 decisions = 486.57549086 tokens/decision. GPU-time cost USD 0.00848657130636103/1,000; old list-price cost USD 0.0440017941773866/1,000; applied min(old list-price cost, GPU-time cost) = USD 0.00848657130636103/1,000. The GPU-time estimate is lower. Estimated, not a bill. Intelligence, Calibration, measured latency and the common cost-item membership are unchanged. Prior prices remain in the archived v1.7 releases.

Measured p50 latency: 0.060 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: 0.27 s (est. incl. network (2x + 150 ms))

Self-hosted model: measured p50 0.060 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 = 0.27 s. Model authors can ask us to time a hosted endpoint instead (on request).

x2 + 0.15 s (assumption, not measured)

evaluator-owned Lium GPU pod, 1x NVIDIA RTX PRO 6000 Blackwell 96GB (the author's card class), the author's own serving bundle (SGLang image pinned by digest + their compatibility server, single GPU via their documented switch), offline (all credentials unset), driven over loopback only by the JevBench client with its loopback guard; label-free inputs; pod destroyed afterwards; HF Jev Decision Index intake, 6-7 Oct 2026

Measured latency over 1,500 decisions: p50 0.060 s · p95 0.077 s

23 decisions over 10 s (23 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.

Breakdowns (not part of the score)

Chance-corrected competence (0 = chance, 100 = perfect) by category on S+P+L1+L2+L3. Raw values, not equated; the same cells feed the compare view.

Capability by subject topic

Rules, policy & law51.3n=1616
Coding & software70.9n=382
Math & numbers20.1n=370
Finance & commerce57.4n=223
Support & operations58.0n=160
Everyday language78.4n=145
Safety & security64.1n=109

Use cases (TypeSafe categories)

Model routing70.1n=524
Legal & compliance60.0n=515
Customer support51.9n=240
E-commerce52.7n=119
Insurance claims57.2n=119
Risk assessment40.5n=118
Financial crime30.0n=101
Feature extraction51.7n=99
Lead generation65.0n=95
Recruiting22.9n=94
Knowledge graphs35.4n=94
LLM guardrails75.3n=93
Moderation61.5n=93
Code linting61.6n=90
Search & retrieval78.0n=90
Science32.5n=90
Advertising25.0n=88
Gaming6.5n=86
Demand forecasting9.3n=83

Cells need at least 30 answered items; † fewer than 30 (indicative only; drawn as hollow, lighter points on the compare radars).

JevBench v1.6.43 values (method v1.6, before the 10 s rule)

Capability 72.41 · Composite 67.38 · rank #42 on the all-systems board · #22 open weights

intelligence 58.67 · calibration 86.14 · speed 91.03 · cost 50.70

v1.6 speed basis (124 items): p50 0.059 s · p95 0.073 s

These figures are kept for comparison. The current figures above are re-scored under the 10 s rule from the recorded per-decision times and replace them.

Published source