Skip to main content

Jared PalmerDecision matchup

Kev 0.8Bvs.Kev 9B

Which model makes better decisions?

Kev 9B does better in independent tests, while Kev 0.8B is more popular.

Independent tests
Kev 0.8B0 of 3
Kev 9Bahead3 of 3
Jev Decision Index, Kevala decision benchmark report, Duan et al., General Decision Models: Benchmarking and Insights Beyond Jev
Popularity
Kev 0.8Bahead45
Kev 9B40
out of 100
Weights and access
Kev 0.8Bopen, local
Kev 9Bopen, local
Released
Kev 0.8BSeptember 20, 2026
Kev 9BSeptember 20, 2026

Pick Kev 0.8B for

more popular

Pick Kev 9B for

Jev Decision IndexKevala decision benchmark reportDuan et al., General Decision Models: Benchmarking and Insights Beyond Jev

Shared tests

Independent evaluators and leaderboards only, the newest shared snapshot per evaluator. With 95 % ranges a lead counts when they do not overlap; for hit rates, when it holds in a 95 % test. Without any uncertainty figure at most “slightly ahead”.

Jev Decision Index

Kev 9B ahead

Kev 9B is clearly ahead in 21 of 37 subtests and slightly in 13, Kev 0.8B in 1. 2 are level.

Home appliances · Decision Index 0.4.1

Kev 0.8B0.0
Kev 9B11.4

88 tasks, as of October 28, 2026

POP909 · Decision Index 0.4.1

Kev 0.8B1.5
Kev 9B10.2

2,000 tasks, as of October 28, 2026

SATA-Bench · Decision Index 0.4.1

Kev 0.8B9.0
Kev 9B30.7

1,650 tasks, as of October 28, 2026

Show all 37 subtests
SubtestKev 0.8BKev 9BVerdict
Home appliances · Decision Index 0.4.10.011.4Kev 9B ahead
POP909 · Decision Index 0.4.11.510.2Kev 9B ahead
SATA-Bench · Decision Index 0.4.19.030.7Kev 9B ahead
GSM8K · Decision Index 0.4.119.054.4Kev 9B ahead
MMLU-Pro · Decision Index 0.4.122.356.5Kev 9B ahead
New Yorker · Decision Index 0.4.127.856.6Kev 9B ahead
HellaSwag · Decision Index 0.4.142.982.7Kev 9B ahead
BBH · Decision Index 0.4.141.466.3Kev 9B ahead
ChessBench · Decision Index 0.4.17.912.0Kev 9B ahead
API-Bank · Decision Index 0.4.143.765.4Kev 9B ahead
CRUXEval · Decision Index 0.4.137.755.4Kev 9B ahead
WinoGrande · Decision Index 0.4.152.875.1Kev 9B ahead
When2Call · Decision Index 0.4.143.260.7Kev 9B ahead
BFCL · Decision Index 0.4.170.595.2Kev 9B ahead
CLadder · Decision Index 0.4.150.563.5Kev 9B ahead
BPoMP · Decision Index 0.4.147.758.9Kev 9B ahead
Habermas · Decision Index 0.4.133.440.0Kev 9B ahead
PhishNChips · Decision Index 0.4.151.761.6Kev 9B ahead
HoVer · Decision Index 0.4.151.060.7Kev 9B ahead
MuSR · Decision Index 0.4.148.956.5Kev 9B ahead
Humicroedit · Decision Index 0.4.150.957.3Kev 9B ahead
cfcolor · Decision Index 0.4.155.453.0Kev 0.8B ahead
CLINC150 · Decision Index 0.4.115.077.1Kev 9B slightly ahead
iSarcasmEval · Decision Index 0.4.119.943.2Kev 9B slightly ahead
Amazon ESCI · Decision Index 0.4.122.949.1Kev 9B slightly ahead
ANLI · Decision Index 0.4.133.658.1Kev 9B slightly ahead
ACOS · Decision Index 0.4.19.917.0Kev 9B slightly ahead
ContractNLI · Decision Index 0.4.140.965.3Kev 9B slightly ahead
RAGTruth · Decision Index 0.4.146.270.6Kev 9B slightly ahead
VAST · Decision Index 0.4.140.656.7Kev 9B slightly ahead
NLI4CT · Decision Index 0.4.155.975.2Kev 9B slightly ahead
FinEntity · Decision Index 0.4.171.489.7Kev 9B slightly ahead
ToolRet · Decision Index 0.4.151.763.8Kev 9B slightly ahead
BRIGHT · Decision Index 0.4.131.838.2Kev 9B slightly ahead
BANKING77 · Decision Index 0.4.177.683.3Kev 9B slightly ahead
HLE · Decision Index 0.4.113.811.0level
GPQA Diamond · Decision Index 0.4.133.237.8level

Source · latest October 28, 2026

Kevala decision benchmark report

Kev 9B ahead

Kev 9B is clearly ahead in 2 of 3 subtests and slightly in 1, Kev 0.8B in none.

Kevala semif-perturbations108 decision accuracy

Kev 0.8B70.7
Kev 9B96.3

95 % ranges 57.4 to 84.0 and 90.7 to 100.0, 972 tasks, as of October 23, 2026

Kevala semif-authored144 decision accuracy

Kev 0.8B71.1
Kev 9B91.9

95 % ranges 64.6 to 77.3 and 88.2 to 95.6, 1,296 tasks, as of October 23, 2026

Kevala kevala-authored36 decision accuracy

Kev 0.8B89.8
Kev 9B96.3

95 % ranges 79.6 to 97.2 and 89.8 to 100.0, 324 tasks, as of October 23, 2026

Source · latest October 23, 2026

Duan et al., General Decision Models: Benchmarking and Insights Beyond Jev

Kev 9B ahead

JEVal aida-testc accuracy

Kev 0.8B38.0 %
Kev 9B96.0 %

200 tasks, as of October 22, 2026

Source · latest October 22, 2026

Vendor figures

Figures from vendor reports that cover both models. They do not count toward the verdict.

Show 1 vendor figures
FigureSourceKev 0.8BKev 9B
14-public-dataset calibration error, per-dataset average (lower is better)Jared Palmer, Introducing Kev0.100.10

All data side by side

Kev 0.8BKev 9B
KindAdapterAdapter
DeveloperJared PalmerJared Palmer
StatusOpen sourceOpen source
ReleasedSeptember 20, 2026September 20, 2026
Weightsopenopen
Runs locallyyesyes
LicenseApache-2.0 for the adapter and headApache-2.0 for the adapter and head
Popularity (out of 100)4540