Laya alternatives
Jev was released in September 2026, and within weeks several open models with the same typed-question interface appeared. Laya is a 322M to 421M encoder. Most alternatives are open LLMs, mostly Qwen, fine-tuned at 0.5B to 35B parameters to score options without generating text. They trade speed for accuracy.
468 projects
kime-bench
tamnd
Benchmarks kime against Laya, laya-mlx, laya-coreml, and Jev on matched hardware
ewm-laya-bonsai-lab
alexmy21
Uses notebooks to inspect a Rust Laya daemon's interactions and compare Laya with Jev and a mock
jev-kev-laya-selfhost
suarify
Provides a Dockerized, self-hosted HTTP API for Laya typed decisions with agent-skill documentation
laya-vs-dgpl-arena
vk-alto-none
Stages AI decision models in a game arena, including local Laya and hosted Jev
laya-zeroshot-agent-guard-eval
Zephyr4772
An evaluation measures zero-shot Laya checkpoints as guards between agent steps
Smart-Support-Ticket-Router-Using-Laya
affan1311
Routes support tickets with Laya and uses Gemini to draft replies, with a benchmark against a Gemini-only pipeline
Smart-Support-Ticket-Classifier-Using-Laya
affanhyder-diggit
Routes support tickets with Laya and benchmarks the results against a Gemini-only pipeline
privatemode-decisions-benchmark
edgelesssys
A benchmark compares speed, accuracy, and cost for Jev, Laya, and Privatemode decisions
jev-equivalent-research
hjl1045
Evaluates Jev, Laya, and a GPT-5.6 Luna baseline on synthetic auto-claims classification
MacJev-322M-4K-Laya-MLX
lawrence3699
Runs the MacJev typed-decision model natively on Apple silicon with MLX and compares it with Laya
jev-bench
brandonrc
Benchmarks Jev, Laya, Claude Haiku, and CLM on package-registry triage tasks
decision-lab
Amine-LG
Provides a visual playground to build decisions and compare Jev, OpenJEV, and local Laya
KVev
AdudodlaVarish
Serves cached long-document chat and uses Laya for offline routing experiments
decision-model-arena
sathik11
Compares Jev, Laya, and Microsoft Foundry for typed decisions in enterprise incident triage
tool-output-pruning-lab
Dymyt-ry
Benchmarks Laya and other selectors for pruning tool output in coding-agent sessions
kime-compat
tamnd
Checks whether kime supports TypeSafe and Laya APIs, clients, SDKs, and tests
narde
anliang0306
Reimplements the Laya decision engine and verifies behavioral parity against upstream
jev-agent-risk-gate
jayeshvpatil
Compares Jev, Claude, fine-tuned Qwen, and open Laya in experiments on agent shell-command risk gates
d3code-calibration
gkastanis
Compares Jev and open-weights Laya probability calibration against human ratings
jev-context-rot
casperkwok
Reproduces an experiment comparing context sensitivity in Jev, Laya and Decider models
jev-vs-laya
Cognition-Forge
Compares Jev and Laya variants across typed-decision tasks
laya-games
BM-Popcorn
Demonstrates Laya-driven turn-based games and fine-tunes the model on engine-labeled positions
laya_bio
maris205
Adapts the Laya encoder into a shared candidate-scoring model for biological sequence tasks
laya_demo
itsvrushabh
Demonstrates and benchmarks Laya’s typed decision engine across interactive use cases
The models
Sizes and licences are from each model's Hugging Face card.
| model | maker | sizes | licence |
|---|---|---|---|
| Kev | Jared Palmer | 0.5B to 27B (benchmarked: 0.8B, 4B, 9B) | Apache 2.0 |
| Decider | Mapika | 0.8B to 35B, plus a 2B vision model | Apache 2.0 |
| Clef, Clef-flash | Cloudflare | 27B, 9B | Apache 2.0 |
| GLiNER2.5-Decide | Fastino | 340M English, 287M multilingual, 1B | Apache 2.0 |
| Julia-1 | Supersonic Labs | 144M | Apache 2.0 |
| Jebadiah | Frontier Infra | 4B, 9B, 27B | Apache 2.0 |
| pplx-decider | Perplexity | ~26B ("27b" in the name) | Apache 2.0 |
- Kev runs a frozen Qwen3.5 base prefill-only, with a LoRA adapter and a pointer head over the options. The card says states of up to 65,536 tokens are served and 8,192 are validated.
- Mapika Decider calls itself an open reproduction of the System One model class. Its training code and data registry are on GitHub.
- Clef and Clef-flash read text, JSON, images and video, per their cards. Cloudflare's announcement says they are Jev-API compatible and also served on Workers AI.
- GLiNER2.5-Decide is a classification-only member of the GLiNER2.5 family: any label set at call time, single- or multi-label.
- Julia-1 is, like Laya, an encoder: it starts from mmBERT-small. See ModernBERT and mmBERT in Laya.
- Jebadiah ships a standalone server that answers
POST /v1/systemoneon one CUDA GPU or an Apple Silicon Mac. - pplx-decider v1.1 uses a separate decision readout instead of a normal language-model head, so its card asks you to use the included inference code.
The hosted reference is Jev from TypeSafe. Its weights are closed and it is billed per input token. The differences from Laya are in Laya vs Jev.
What independent benchmarks found
akitaonrails (October 10, 2026) used synthetic, AI-assisted cases in English and Brazilian Portuguese without independent human labels. Chance was about 31%. Local models ran on a Ryzen AI Max+ 395. Compact-set accuracy and median latency:
| model | accuracy | latency |
|---|---|---|
| Jev (hosted) | 97.6% | 255 ms |
| Clef (hosted) | 95.8% | 302 ms |
| Perplexity Decider 1.1 | 99.4% | 1,547 ms |
| Jebadiah 4B v2 | 90.5% | 205 ms |
| Kev 9B | 88.1% | 561 ms |
| Kev 4B | 87.5% | 246 ms |
| Mapika Decider | 87.5% | 186 ms |
| Clef-flash (local) | 87.5% | 618 ms |
| Kev 0.8B | 63.1% | 86 ms |
| GLiNER2.5-Decide (English only) | 52.4% | 208 ms |
| Laya (English) | 42.9% | 41 ms |
| Julia-1 | 31.5% | 12 ms (GPU) |
The author's conclusion: if Jev-like accuracy is a requirement, Laya, Julia-1 and the 0.6B Decision 3.0 Edge are not usable out of the box. He also notes that judging these small encoders by the accuracy column alone misreads them: they are built to be fine-tuned and run at the edge. Fine-tuning was not tested.
Arize (October 2026) ran a bracket of hallucination detection on RAGTruth, LLM-evaluator suites and agent routing. Balanced accuracy on RAGTruth was 85.5% for Jev, 85.2% for Clef 27B, 76.8% for Kev 9B and 59.8% for laya-multilingual. Laya was the fastest model at 0.04 s p50 on an M4 Max, and had the highest calibration error (0.330).
systemonemodels.ai lists 29 open Jev alternatives with the accuracy each publisher reports. Those figures come from different test sets, so they cannot be compared with each other.
How to choose
- Speed, CPU or on-device: the encoders. Laya and Julia-1 answer in tens of milliseconds, and Laya has the most ports, from Mac to the browser. Plan to fine-tune or to keep a fallback for hard cases.
- Accuracy without training: the 4B and larger models. They need a GPU or an Apple Silicon Mac with enough memory, and they are several times slower.
- No servers: Jev, Clef on Workers AI, or a hosted Laya API.
Whichever you pick, run a few hundred of your own labelled examples through two or three candidates. The rankings above come from small or synthetic sets, and their authors say so.
More ways to use Laya