whatmodel.sh / decision / laya (DecisionBench 1c5edc17)
laya (DecisionBench 1c5edc17)
↗ share
+ contribute
laya (DecisionBench 1c5edc17) is an open-weights decision model from ConvAI Innovations. Its published size is 0.42129383B parameters. Licence: apache-2.0.
Catalogue identity notes
DecisionBench v1 comparison: convaiinnovations/laya, revision 1c5edc17a7acd8701df6fc341c0d179f1c62c982; decision-model; adapter laya-hf; probability source native-decisions-probability-distribution. Kept separate from newer or unpinned Decision Index recipes; standard prompt, not compact. No unverified OpenRouter price imported. Current repository metadata checked separately; an unpinned alias is not evidence that today’s weights reproduce this historical run. Reviewed Ollaya local route added 2026-10-31: laya:en, manifest c305a9276531a47000bf93559d2c94f1ed6cbb67055c9151084682dad7655e9d, matching the independent accuracy-run manifest. Native family implementation; published runtime precision label F16. Memory is measured peak GPU increase (1654 MiB, rounded upward to 1.7 GiB) on RTX 5090 during the short five-question triage workload, not download size or a maximum-context guarantee. Runtime installation is required; no fresh execution performed here. Historical benchmark adapter and input protocol still vary by result.
Model information sources
· Weights licence ↗ · Parameter count ↗
Published served parameter count; not a RAM or download-size measurement.
$ how it compares
The decision models ranked either side of laya (DecisionBench 1c5edc17), out of 357 with a score, then more from ConvAI Innovations.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 329 |  Von 1.3 | 25±1 | model files | ||
| 330 |  Bosun v3.1 0.6B LoRA | 25±1 | model files | ||
| 331 | laya (DecisionBench 1c5edc17) | 25±1 | local | ||
| 332 |  kev-0.5b (DecisionBench 9ce2fd39) | 24±2 | model files | ||
| 333 | Crivo-17M | 24±1 | model files | ||
| more from ConvAI Innovations | |||||
| 315 | laya-typed-decisions (DecisionBench bc76315b) | 29±2 | model files | ||
| 340 | Laya | 21±1 | model files | ||
see all decision models · find the ones that fit your machine
$ what it's good at
knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #193 of 263 · 11% of its score · 9 tests
Beats StartLux-Decision v1.1 0.8B, Sieve 2B v3, OneJev1.6-0.8B here, though they rank higher overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Ollaya · BoolQ | 83.7% ✎ | ▁··▃▃▁▄▁▃█▇▄ beats 11/25 | 3.5% |
| Ollaya · PubMedQA | 50.0% ✎ | ▃▂▃▃·▂▃▃▆█▆▂ beats 5/25 | 3.2% |
| Ollaya · SQuAD 2 answerability | 61.9% ✎ | ▁▃▃▁▇█▃▃▄▃·▁ beats 1/25 | 1.6% |
| Ollaya · VitaminC verification | 79.0% ✎ | ▁···▃·▄█▃··▁ beats 18/25 | 1.4% |
| Ollaya · MultiNLI | 87.6% ✎ | ▂··▄██▂▆▆█▆▂ beats 13/25 | 1.0% |
| system1bench · boolq | 84.6% ✎ | █··········█ beats 0/1 | — |
| system1bench · xnli en | 86.0% ✎ | █··········█ beats 1/1 | — |
| system1bench · xnli zh | 61.5% ✎ | █··········█ beats 0/1 | — |
1 more test
| test | result | vs the others in its run | counts |
|---|---|---|---|
| system1bench · turtlebench | 42.6% ✎ | █··········█ beats 0/1 | — |
workflows▁▁▁▃▁▂▂▆█▄▁▁ #102 of 114 · 88% of its score · 58 tests
Trails GLiNER2.5-Decide here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| DecisionBench v1 · action selection (all-request accuracy) submitted | 25.7% ✎ | ▂▁▂▃▄▃▁▁▂█▇▄ beats 11/56 | 10% |
| DecisionBench v1 · routing (all-request accuracy) submitted | 30.0% ✎ | ▁▁▁▁▃▂▁▁▁▁▃█ beats 12/56 | 8.4% |
| DecisionBench v1 · completion assessment (all-request accuracy) submitted | 35.0% ✎ | ▁▁▃▃▄·▂▁▁▆█▂ beats 5/56 | 7.6% |
| DecisionBench v1 · bounded extraction (all-request accuracy) submitted | 24.8% ✎ | ▃▃▂·▁▁▁▁▁▁█▅ beats 7/56 | 6.3% |
| DecisionBench v1 · navigation (all-request accuracy) submitted | 15.0% ✎ | ▁▁·▃▁▁▁▁▁▁▃█ beats 6/56 | 5.2% |
| DecisionBench v1 · triage (all-request accuracy) submitted | 41.0% ✎ | ▁·▁·▂▃▄▃▂█▄▂ beats 10/56 | 4.9% |
| Ollaya · Typed decisions test | 35.8% ✎ | ▄·▂▆▄▄▂·▄█▆▆ beats 1/22 | 4.6% |
| Ollaya · MASSIVE German intent | 40.9% ✎ | ▁····▃·▂▃▆█▃ beats 1/23 | 4.3% |
50 more tests
$ where to run it
Other ways to run it
- Run it as published · needs 1.7 GB, NVIDIA only · from github.com/ollaya-dev
ollaya run laya:en --preset triage "The customer requests a refund."
Download model files ↗
Published model files verified. This is the revision recorded by the benchmark. A matching Ollaya runtime manifest and published short-workload GPU memory measurement were reviewed separately; see local variants and identity_note. The command was not executed in this review. Source · checked 2026-10-31.
Model repository and documentation: huggingface.co/convaiinnovations ↗
$ how the score was worked out · Hanno Labs DecisionBench v1 r… 66%, Ollaya · original runs 33%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 Hanno Labs DecisionBench v1 reviewed results · submitted; independent execution unverified | 27 | 66% |
| Ollaya · original runs | 14 | 33% |
| prior · pull toward the middle | 1% | |
| weighted average | = 24.5 |
not counted (26)
| CYMCharming · system1bench | system1bench · ag news [correct] | 94.3% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · emotion [correct] | 59.1% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · banking77 [correct] | 55.3% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · boolq [correct] | 84.6% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · sst5 [correct] | 34.6% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · xnli en [correct] | 86.0% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · xnli zh [correct] | 61.5% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · massive en [correct] | 54.1% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · massive zh [correct] | 30.4% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · prompt injections [correct] | 70.7% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · typed decisions [correct] | 36.4% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jevbench original [correct] | 70.8% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jevbench easy [correct] | 95.8% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jevbench hard [correct] | 29.7% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · reflexbench reflex-public-choice-v1 [correct] | 59.0% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jev laya triage [correct] | 62.5% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jev laya moderation [correct] | 67.8% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jev laya routing [correct] | 64.2% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jev laya claims [correct] | 90.0% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jev laya reviews [correct] | 70.7% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jev laya guard [correct] | 60.6% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jev laya multilingual [correct] | 57.8% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · jev laya needle [correct] | 50.2% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · clinc150 oos [correct] | 55.7% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · turtlebench [correct] | 42.6% | fewer than three models have this test yet, or this source ran it for this model alone |
| CYMCharming · system1bench | system1bench · aegis2 prompt [correct] | 49.6% | fewer than three models have this test yet, or this source ran it for this model alone |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json