Convai InnovationsDecision modelOpen source
Laya Typed Decisions
PopularityRank 271 of 876
40out of 100same as last month
Summary
Laya Typed Decisions is Convai’s variant for agent-trace observability, customer service, invoice processing, and security incidents. It uses 421 million parameters, a 1,024-token context, and local non-autoregressive readouts. The result remains a Choice, Noul, or Score with probabilities.
Those four use cases distinguish Laya Typed Decisions from Laya for general English input and Laya Multilingual for multiple languages.
All benchmark results
Every published result with its source, test setup and comparison models.
Show all 179 published benchmark results
Page 1 of 15
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
Zero-shot IE multilingual: popular-language tier accuracy (Show measurement, test conditions, and source)
- Source value
- 100
- Score
- 100
- Metric
- accuracy, source table rounded to whole percent
- Unit
- %
- Benchmark version
- Zero-shot IE multilingual snapshot at repository commit c45a00c25d2731ac7047def97b91421aaff2f611
- Category
- multilingual-popular
- Direction
- Higher is better
- Sample
- 18
- Source type
- independent-evaluator
- Evaluator
- Zero-shot IE benchmark README, pinned commit c45a00c25d2731ac7047def97b91421aaff2f611
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. The pinned author README's multilingual table reports the Popular tier for exact result row “Laya typed-decisions”: 100%. The task comprises six sentiment texts per each of nine languages; this tier includes Spanish, French and Simplified Chinese, 18 texts total. The result cell is taken directly from the pinned README table (body SHA-256 7905d1cbfb4d75d65d02a5e6c537e1e84fe8b5c388f33075a6c0db42fa0cac98), not recomputed from the separate language scores. The public runner describes blind back-translation; it uses label options with English descriptions. No immutable model/API revision or confidence interval is published.
- Context
- The pinned author README publishes rounded whole-percent tier cells; these exact displayed cells are used, not recomputed from the language-level data. This is a custom, small evaluation rather than a standardized leaderboard. The author reports blind back-translation and says the full text table was in a PR for native-speaker review; this review itself was not independently confirmed. System/model revisions and full decoding settings are not pinned.
Zero-shot IE multilingual: medium-resource-language tier accuracy (Show measurement, test conditions, and source)
- Source value
- 44
- Score
- 44
- Metric
- accuracy, source table rounded to whole percent
- Unit
- %
- Benchmark version
- Zero-shot IE multilingual snapshot at repository commit c45a00c25d2731ac7047def97b91421aaff2f611
- Category
- multilingual-medium
- Direction
- Higher is better
- Sample
- 18
- Source type
- independent-evaluator
- Evaluator
- Zero-shot IE benchmark README, pinned commit c45a00c25d2731ac7047def97b91421aaff2f611
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. The pinned author README's multilingual table reports the Medium tier for exact result row “Laya typed-decisions”: 44%. The task comprises six sentiment texts per each of nine languages; this tier includes Vietnamese, Turkish and Ukrainian, 18 texts total. The result cell is taken directly from the pinned README table (body SHA-256 7905d1cbfb4d75d65d02a5e6c537e1e84fe8b5c388f33075a6c0db42fa0cac98), not recomputed from the separate language scores. The public runner describes blind back-translation; it uses label options with English descriptions. No immutable model/API revision or confidence interval is published.
- Context
- The pinned author README publishes rounded whole-percent tier cells; these exact displayed cells are used, not recomputed from the language-level data. This is a custom, small evaluation rather than a standardized leaderboard. The author reports blind back-translation and says the full text table was in a PR for native-speaker review; this review itself was not independently confirmed. System/model revisions and full decoding settings are not pinned.
Zero-shot IE multilingual: rare-language tier accuracy (Show measurement, test conditions, and source)
- Source value
- 33
- Score
- 33
- Metric
- accuracy, source table rounded to whole percent
- Unit
- %
- Benchmark version
- Zero-shot IE multilingual snapshot at repository commit c45a00c25d2731ac7047def97b91421aaff2f611
- Category
- multilingual-rare
- Direction
- Higher is better
- Sample
- 18
- Source type
- independent-evaluator
- Evaluator
- Zero-shot IE benchmark README, pinned commit c45a00c25d2731ac7047def97b91421aaff2f611
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. The pinned author README's multilingual table reports the Rare tier for exact result row “Laya typed-decisions”: 33%. The task comprises six sentiment texts per each of nine languages; this tier includes Sinhala, Icelandic and Welsh, 18 texts total. The result cell is taken directly from the pinned README table (body SHA-256 7905d1cbfb4d75d65d02a5e6c537e1e84fe8b5c388f33075a6c0db42fa0cac98), not recomputed from the separate language scores. The public runner describes blind back-translation; it uses label options with English descriptions. No immutable model/API revision or confidence interval is published.
- Context
- The pinned author README publishes rounded whole-percent tier cells; these exact displayed cells are used, not recomputed from the language-level data. This is a custom, small evaluation rather than a standardized leaderboard. The author reports blind back-translation and says the full text table was in a PR for native-speaker review; this review itself was not independently confirmed. System/model revisions and full decoding settings are not pinned.
StreamDecisionBench (Show measurement, test conditions, and source)
- Source value
- 1.46
- Score
- 1.46
- Metric
- Mean log-AUC 0.5–8 s
- Unit
- %
- Direction
- No direction published
- Sample
- 480
- Source type
- independent-evaluator
- Evaluator
- StreamDecisionBench GitHub README
- Status
- active
- Retrieved at
- 2026-10-18
- Methodology
- Independent evaluation. Self-hosted open-weight setting, exact source row “Laya typed-decisions”. Three passes over all 480 states per pass (8 scenarios in 4 families), sampled at a 2 s step; the source reports means. The README says all 17 self-hosted settings use BF16 backbones, each on one RTX PRO 6000 (96 GB), with benchmark and model on the same GPU host. Same-host latency includes request processing, runtime queueing, and inference, and excludes download, initialization, and warmup. Serving stack, exact checkpoint revision, and other runtime settings are not stated.
StreamDecisionBench (Show measurement, test conditions, and source)
- Source value
- 1.46
- Score
- 1.46
- Metric
- Mean untimed (%)
- Unit
- %
- Direction
- No direction published
- Sample
- 480
- Source type
- independent-evaluator
- Evaluator
- StreamDecisionBench GitHub README
- Status
- active
- Retrieved at
- 2026-10-18
- Methodology
- Independent evaluation. Self-hosted open-weight setting, exact source row “Laya typed-decisions”. Three passes over all 480 states per pass (8 scenarios in 4 families), sampled at a 2 s step; the source reports means. The README says all 17 self-hosted settings use BF16 backbones, each on one RTX PRO 6000 (96 GB), with benchmark and model on the same GPU host. Same-host latency includes request processing, runtime queueing, and inference, and excludes download, initialization, and warmup. Serving stack, exact checkpoint revision, and other runtime settings are not stated.
Typed Decision Candidate Coverage (TREC 5 Nat.) (Show measurement, test conditions, and source)
- Source value
- 97.2
- Score
- 97.2
- Metric
- Missing-answer detection rate D
- Unit
- %
- Benchmark version
- arXiv:2610.03387v1
- Category
- typed-decision
- Direction
- Higher is better
- Sample
- 109
- Lower confidence bound
- 92.2
- Upper confidence bound
- 99.1
- Source type
- independent-evaluator
- Evaluator
- Lu and Wu, Benchmarking Candidate Coverage in Typed Decision Models
- Status
- active
- Retrieved at
- 2026-10-20
- Methodology
- Independent evaluation. TREC with five natural-name candidates (TREC 5 Nat.), 109 test texts; D measures detection of missing reference answers and F false rejection of present candidates. The paper reports marginal Wilson 95% intervals. Tested: Laya 0.3.21 without fine-tuning, English checkpoint 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; A40, PyTorch 2.8.0, Transformers 4.56.2, batch 8, sequence/head budgets 512/256, BF16.
- Context
- The paper describes the study as exploratory, not an independent preregistered confirmation; read the 97.2% detection together with the 69.7% false-rejection rate. Do not generalize to task, prompt, seed, or hosted-model variability.
Typed Decision Candidate Coverage (TREC 5 Nat.) (Show measurement, test conditions, and source)
- Source value
- 69.7
- Score
- 69.7
- Metric
- Present-control false-rejection rate F
- Unit
- %
- Benchmark version
- arXiv:2610.03387v1
- Category
- typed-decision
- Direction
- Lower is better
- Sample
- 109
- Lower confidence bound
- 60.5
- Upper confidence bound
- 77.6
- Source type
- independent-evaluator
- Evaluator
- Lu and Wu, Benchmarking Candidate Coverage in Typed Decision Models
- Status
- active
- Retrieved at
- 2026-10-20
- Methodology
- Independent evaluation. TREC with five natural-name candidates (TREC 5 Nat.), 109 test texts; D measures detection of missing reference answers and F false rejection of present candidates. The paper reports marginal Wilson 95% intervals. Tested: Laya 0.3.21 without fine-tuning, English checkpoint 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; A40, PyTorch 2.8.0, Transformers 4.56.2, batch 8, sequence/head budgets 512/256, BF16.
- Context
- The paper describes the study as exploratory, not an independent preregistered confirmation; read the 97.2% detection together with the 69.7% false-rejection rate. Do not generalize to task, prompt, seed, or hosted-model variability.
Laya Typed-Decisions E7 option-order prediction-flip rate (Show measurement, test conditions, and source)
- Source value
- 6.75
- Score
- 6.75
- Metric
- prediction flip rate under option-order perturbation
- Unit
- %
- Benchmark version
- arXiv:2609.33843v1, E7
- Category
- decision-robustness
- Direction
- No direction published
- Source type
- independent-evaluator
- Evaluator
- Nandakishore, Laya as a Typed Probabilistic Assessor: An Independent Reproduction and a Preregistered Study of Calibration and Selective Escalation
- Status
- active
- Retrieved at
- 2026-10-22
- Methodology
- Independent evaluation. E7 tests option-order sensitivity on calibration cases: 300 choice decisions were evaluated under four seeded option orders. The source provides no immutable evaluated-weight hash.
- Context
- This result is exploratory and uses only the calibration split, which the source describes as half of the official test split; it is not an independent holdout. The source reports 300 decisions and four seeded permutations but does not clearly define the denominator of the 6.75% flip rate. No confidence interval or immutable weight revision is provided.
DecideBench overall decision accuracy (Show measurement, test conditions, and source)
- Source value
- 59.8
- Score
- 59.8
- Metric
- accuracy
- Unit
- %
- Benchmark version
- DecideBench results/v1
- Direction
- Higher is better
- Sample
- 400
- Source type
- community-reported
- Evaluator
- DecideBench
- Status
- active
- Retrieved at
- 2026-10-22
- Methodology
- Community-reported result. Laya typed-decisions 421M was evaluated self-hosted on an NVIDIA L4 and runs zero-shot in this comparison. The benchmark contains 400 decisions in 200 contrastive pairs. Study/result period: results reported for 2026-09-28 to 2026-10-09
- Context
- The report names the model and size but does not bind an immutable weight revision for this run.
DecideBench pair accuracy (Show measurement, test conditions, and source)
- Source value
- 36.5
- Score
- 36.5
- Metric
- pair accuracy
- Unit
- %
- Benchmark version
- DecideBench results/v1
- Direction
- Higher is better
- Sample
- 200
- Source type
- community-reported
- Evaluator
- DecideBench
- Status
- active
- Retrieved at
- 2026-10-22
- Methodology
- Community-reported result. Laya typed-decisions 421M was evaluated self-hosted on an NVIDIA L4 and runs zero-shot in this comparison. The benchmark contains 400 decisions in 200 contrastive pairs. Study/result period: results reported for 2026-09-28 to 2026-10-09
- Context
- The report names the model and size but does not bind an immutable weight revision for this run.
Jev docs bilingual v2 Compact EN accuracy (Show measurement, test conditions, and source)
- Source value
- 50
- Score
- 50
- Metric
- Compact EN accuracy
- Unit
- %
- Benchmark version
- Jev docs bilingual expansion v2, Compact condition
- Category
- decision
- Direction
- Higher is better
- Sample
- 84
- Source type
- community-reported
- Evaluator
- akitaonrails/jev-docs-benchmarks, bilingual v2 results
- Status
- active
- Retrieved at
- 2026-10-23
- Methodology
- Community-reported result. Bilingual expansion v2, Compact condition; source reports 84 authored families with paired translations and three correlated conditions. The model row is `laya-typed`; target checkpoint revision is not stated.
- Context
- The source says 84 authored families; it does not separately report model-specific completion counts or checkpoint revision. Task cells have only 12 families and are exploratory.
Jev docs bilingual v2 Compact PT-BR accuracy (Show measurement, test conditions, and source)
- Source value
- 35.7
- Score
- 35.7
- Metric
- Compact PT-BR accuracy
- Unit
- %
- Benchmark version
- Jev docs bilingual expansion v2, Compact condition
- Category
- decision
- Direction
- Higher is better
- Sample
- 84
- Source type
- community-reported
- Evaluator
- akitaonrails/jev-docs-benchmarks, bilingual v2 results
- Status
- active
- Retrieved at
- 2026-10-23
- Methodology
- Community-reported result. Bilingual expansion v2, Compact condition; source reports 84 authored families with paired translations and three correlated conditions. The model row is `laya-typed`; target checkpoint revision is not stated.
- Context
- The source says 84 authored families; it does not separately report model-specific completion counts or checkpoint revision. Task cells have only 12 families and are exploratory.
Page 1 of 15
Popularity compared
How often the model is searched for, used and mentioned, compared with every model of its category.
Popularity index
0 to 100, higher is more popular
20 of 37 models
As of November 2026 · Source: gradually.ai popularity index
Over time
Weekly values from October 18, 2026 to November 6, 2026
Laya Typed DecisionsIntern-Decision-4B 40Kev 9B 40pplx-decider-v1-27b 39Decider 0.8B 39
Method changes in the timeline, a jump at these points is no market event:
- 1November 1, 2026: Value only with search or usage data, 30-day attention
What is gaining popularity right now, across models, vendors, harnesses and AI subscriptions, is collected in All AI trends.
Laya Typed Decisions compared
- 1Laya Typed Decisions vs. Clef-Flash1 shared test
- 2Laya Typed Decisions vs. GLiNER2.5-Decide4 shared tests
- 3Laya Typed Decisions vs. Intern-Decision-4B4 shared tests
- 4Laya Typed Decisions vs. Kev 0.8B4 shared tests
- 5Laya Typed Decisions vs. Decider 0.8B4 shared tests
- 6Laya Typed Decisions vs. Intern-Decision-0.8B4 shared tests
- 7Laya Typed Decisions vs. Intern-Decision-2B4 shared tests
- 8Laya Typed Decisions vs. GLiDE1 shared test
- 9Laya Typed Decisions vs. Laya1 shared test
Similar models
More models of the same kind as Laya Typed Decisions, chosen by model line, developer, size, price and release date.
- Laya MultilingualConvai Innovations · same developer
- LayaConvai Innovations · same developer
- GLiNER2.5-DecideFastino Labs · similar size
- GLiNER2.5-multi-DecideFastino Labs · similar size
- Von 1.2Von · similar size
- NanoJevTianyuCodings · similar size
- Decider 0.8BMapika · similar size
- Intern-Decision-0.8BInternLM · similar size
All specifications
Architecture
Input and output
Watermarking and provenance
AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.
- AI watermarking
- Unknown
- no source
- Watermarking technique
- Not known
- no source
Licence and access
API
Further details
- Notes
- The checkpoint is listed as a separate typed-decisions variant in the official repository.
- Source: Notes
Sources and data date
Show all 28 sources
Benchmark sources at a glance
Which independent evaluators tested this model. This is coverage, not a quality ranking.
No active result from the reviewed external evaluators can currently be attributed to this model. Other source evidence remains available below.
Without vendor claims and historical or disputed results; those stay labelled in the single values.