Skip to main content

Convai InnovationsFastino LabsDecision matchup

Layavs.GLiNER2.5-Decide

Which model makes better decisions?

GLiNER2.5-Decide does better in independent tests.

Independent tests
Laya0 of 3
GLiNER2.5-Decideahead1 of 3
Jev Decision Index, Niklas Heringer, System One AI for Cybersecurity: Not Every Decision Needs a Generative Model, Zero-shot IE benchmark
Popularity
Laya0
GLiNER2.5-Decideahead56
out of 100
Weights and access
Layaopen, local
GLiNER2.5-Decideopen, local
Size
Laya0.42 B
GLiNER2.5-Decide0.34 B
Parameters

Pick Laya for

No clear edge

Pick GLiNER2.5-Decide for

Niklas Heringer, System One AI for Cybersecurity: Not Every Decision Needs a Generative Modelmore popular

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

depends on the subtest

Laya is clearly ahead in 5 of 37 subtests and slightly in 6, GLiNER2.5-Decide in 6 clearly and slightly in 5. 15 are level.

POP909 · Decision Index 0.4.1

Laya5.1
GLiNER2.5-Decide1.7

2,000 tasks, as of October 28, 2026

When2Call · Decision Index 0.4.1

Laya11.9
GLiNER2.5-Decide33.4

3,652 tasks, as of October 28, 2026

API-Bank · Decision Index 0.4.1

Laya11.5
GLiNER2.5-Decide7.7

508 tasks, as of October 28, 2026

Show all 37 subtests
SubtestLayaGLiNER2.5-DecideVerdict
POP909 · Decision Index 0.4.15.11.7Laya ahead
When2Call · Decision Index 0.4.111.933.4GLiNER2.5-Decide ahead
API-Bank · Decision Index 0.4.111.57.7Laya ahead
Habermas · Decision Index 0.4.133.445.5GLiNER2.5-Decide ahead
BFCL · Decision Index 0.4.138.146.2GLiNER2.5-Decide ahead
MMLU-Pro · Decision Index 0.4.113.616.1GLiNER2.5-Decide ahead
BPoMP · Decision Index 0.4.151.658.2GLiNER2.5-Decide ahead
HoVer · Decision Index 0.4.155.850.7Laya ahead
BBH · Decision Index 0.4.134.131.1Laya ahead
Humicroedit · Decision Index 0.4.147.250.6GLiNER2.5-Decide ahead
CLadder · Decision Index 0.4.152.950.1Laya ahead
ACOS · Decision Index 0.4.13.50.0Laya slightly ahead
CLINC150 · Decision Index 0.4.13.260.4GLiNER2.5-Decide slightly ahead
BANKING77 · Decision Index 0.4.114.365.6GLiNER2.5-Decide slightly ahead
ToolRet · Decision Index 0.4.112.838.2GLiNER2.5-Decide slightly ahead
iSarcasmEval · Decision Index 0.4.114.932.4GLiNER2.5-Decide slightly ahead
RAGTruth · Decision Index 0.4.148.825.3Laya slightly ahead
ANLI · Decision Index 0.4.148.730.9Laya slightly ahead
ContractNLI · Decision Index 0.4.129.021.8Laya slightly ahead
VAST · Decision Index 0.4.140.535.3Laya slightly ahead
FinEntity · Decision Index 0.4.161.066.2GLiNER2.5-Decide slightly ahead
Amazon ESCI · Decision Index 0.4.124.423.3Laya slightly ahead
SATA-Bench · Decision Index 0.4.10.30.7level
GPQA Diamond · Decision Index 0.4.127.621.4level
HLE · Decision Index 0.4.114.016.8level
New Yorker · Decision Index 0.4.127.132.2level
ChessBench · Decision Index 0.4.17.78.1level
MuSR · Decision Index 0.4.143.245.2level
HellaSwag · Decision Index 0.4.133.132.2level
GSM8K · Decision Index 0.4.116.917.3level
cfcolor · Decision Index 0.4.152.353.4level
BRIGHT · Decision Index 0.4.119.920.1level
NLI4CT · Decision Index 0.4.147.748.0level
WinoGrande · Decision Index 0.4.150.550.7level
PhishNChips · Decision Index 0.4.150.150.0level
CRUXEval · Decision Index 0.4.140.240.2level
Home appliances · Decision Index 0.4.10.00.0level

Source · latest October 28, 2026

Niklas Heringer, System One AI for Cybersecurity: Not Every Decision Needs a Generative Model

GLiNER2.5-Decide ahead

System One cybersecurity HTTP triage, hard-tier zero-shot accuracy

Laya37.7 %
GLiNER2.5-Decide49.3 %

300 tasks, as of October 18, 2026

Source · latest October 18, 2026

Zero-shot IE benchmark

level

Zero-shot IE spectrum: mixed-pool classification accuracy

Laya68.8
GLiNER2.5-Decide85.4

48 tasks, as of October 17, 2026

Source · latest October 17, 2026

All data side by side

LayaGLiNER2.5-Decide
KindStandaloneClassifier
DeveloperConvai InnovationsFastino Labs
StatusOpen sourceOpen source
Weightsopenopen
Runs locallyyesyes
Parameters0.42 B0.34 B
LicenseApache-2.0Apache-2.0
Popularity (out of 100)056