Myles-G 12B (Launchables)

Data: JevBench v1.6.39, published 2026-10-22

Capability 70.7 · API offering, ranked on the API leaderboard

Last measured: 2026-10-21 · measurement release: v1.6.38

Proprietary vendor-hosted API; weights and serving code not independently inspected

Base model: google/gemma-4-12B-it (self-reported) · Customer states merged LoRA on Gemma 4 12B; not independently verified.

intelligence

57.3

calibration

84.0

speed

85.6

cost

57.4

Composite (secondary)

68.4 · Not ranked on this board

Cost and measurement conditions

$0.026 per 1,000 decisions (estimated)

ESTIMATE, nearest-listed-sibling reference: declared base google/gemma-4-12B-it is UNLISTED in frozen pricing_v15 (sha256 a8a05890ddab2d39d1a1bd543f2b5c83842bdc70aa54766788bc89b7a8a448d7); no exact base reference exists. Same-base published precedents arcbright-elf-1-12b, chon-1.0-12b and metask_jev_rain_12b_trained use google/gemma-3-12b-it, USD 0.05/M input and 0.15/M output. This row uses that same sibling ESTIMATE, not a vendor bill or quote. Launchables list price GBP 0.0015 per million input tokens, output free, is shown for information and is not used for the Cost axis. Applied to provider-reported usage. COMMON COST BASIS (v1.6.30): common 1,485 items, excluding the 15 longest for every row; 779,294 input / 1,485 output tokens. All-item usage 989,607 input / 1,500 output over 1,500 items.

p50 latency: 0.5 s

none (API)

Paid public fast-lane; full S1200+P300 draw v1.6-regular-20261012, unchanged evaluator API runner from Sandy, customer API key; full-set exposure gate logged all sealed items before the first scored request. Native probabilities; latency has no API adjustment.

Published source