Convai InnovationsDecision modelOpen source

Laya

Popularityhardly any signal

under 1out of 100same as last month

Summary

Laya is Convai Innovations’ non-autoregressive System One model for English, structured decisions. It can process text, emails, tickets, or JSON together with typed questions and returns Choice, Noul, and Score with probabilities; for Noul, that value is the probability of yes. The local checkpoint has 421 million parameters and a short 512-token context. It does not generate free-form answers.

Convai publishes Laya Multilingual for multiple languages and Laya Typed Decisions for the named agent-trace, customer-service, invoice, and security data as separate checkpoints.

All benchmark results

Every published result with its source, test setup and comparison models.

Show all 319 published benchmark results

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.

Emotion k=5 local none-score threshold: missing-answer detection (Show measurement, test conditions, and source)

Source value: 19.27710843373494
Score: 19.27710843373494
Metric: missing-answer detection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; confirmation core; calibrated none-probability threshold; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Higher is better
Sample: 166
Lower confidence bound: 13.253012048192772
Upper confidence bound: 25.301204819277107
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; natural Emotion labels; k=5; none-probability threshold 0.9801 selected on 120 calibration texts under the <=5% false-rejection constraint; held-out test metric N=166; 2,000-text bootstrap CI.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw confirmation metrics

Emotion k=5 local none-score threshold: covered-input false rejection (Show measurement, test conditions, and source)

Source value: 7.83132530120482
Score: 7.83132530120482
Metric: covered-input false rejection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; confirmation core; calibrated none-probability threshold; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Lower is better
Sample: 166
Lower confidence bound: 4.216867469879518
Upper confidence bound: 12.048192771084338
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; natural Emotion labels; k=5; none-probability threshold 0.9801 selected on 120 calibration texts under the <=5% false-rejection constraint; held-out test metric N=166; 2,000-text bootstrap CI.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw confirmation metrics

CLINC150 artificial k=5: missing-answer detection (Show measurement, test conditions, and source)

Source value: 86
Score: 86
Metric: missing-answer detection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 artificial k=5; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Higher is better
Sample: 300
Lower confidence bound: 82
Upper confidence bound: 89.66666666666666
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; reference intent artificially removed from a matched 5-intent menu; population: in-scope queries whose reference intent is absent from the menu; exact held-out denominator N=300; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

CLINC150 artificial k=5: covered-input false rejection (Show measurement, test conditions, and source)

Source value: 27.666666666666668
Score: 27.666666666666668
Metric: covered-input false rejection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 artificial k=5; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Lower is better
Sample: 300
Lower confidence bound: 22.333333333333332
Upper confidence bound: 32.666666666666664
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; reference intent artificially removed from a matched 5-intent menu; population: reference-present test queries; exact held-out denominator N=300; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

CLINC150 natural retrieval k=2: missing-answer detection (Show measurement, test conditions, and source)

Source value: 90.32258064516128
Score: 90.32258064516128
Metric: missing-answer detection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 natural retrieval k=2; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Higher is better
Sample: 31
Lower confidence bound: 77.41935483870968
Upper confidence bound: 100
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; fixed BM25 retrieval at k=2; natural candidate menu, without gold-label repair; population: in-scope queries whose reference intent is absent from the menu; exact held-out denominator N=31; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

CLINC150 natural retrieval k=2: covered-input false rejection (Show measurement, test conditions, and source)

Source value: 15.241635687732341
Score: 15.241635687732341
Metric: covered-input false rejection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 natural retrieval k=2; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Lower is better
Sample: 269
Lower confidence bound: 11.152416356877323
Upper confidence bound: 19.702602230483272
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; fixed BM25 retrieval at k=2; natural candidate menu, without gold-label repair; population: reference-present test queries; exact held-out denominator N=269; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

CLINC150 natural retrieval k=2: global out-of-scope detection (Show measurement, test conditions, and source)

Source value: 83
Score: 83
Metric: global out-of-scope detection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 natural retrieval k=2; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Higher is better
Sample: 100
Lower confidence bound: 75
Upper confidence bound: 90
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; fixed BM25 retrieval at k=2; natural candidate menu, without gold-label repair; population: public global out-of-scope test queries; exact held-out denominator N=100; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

CLINC150 natural retrieval k=5: missing-answer detection (Show measurement, test conditions, and source)

Source value: 90.9090909090909
Score: 90.9090909090909
Metric: missing-answer detection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 natural retrieval k=5; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Higher is better
Sample: 11
Lower confidence bound: 72.72727272727273
Upper confidence bound: 100
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; fixed BM25 retrieval at k=5; natural candidate menu, without gold-label repair; population: in-scope queries whose reference intent is absent from the menu; exact held-out denominator N=11; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

CLINC150 natural retrieval k=5: covered-input false rejection (Show measurement, test conditions, and source)

Source value: 30.44982698961938
Score: 30.44982698961938
Metric: covered-input false rejection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 natural retrieval k=5; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Lower is better
Sample: 289
Lower confidence bound: 25.259515570934255
Upper confidence bound: 35.98615916955018
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; fixed BM25 retrieval at k=5; natural candidate menu, without gold-label repair; population: reference-present test queries; exact held-out denominator N=289; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

CLINC150 natural retrieval k=5: global out-of-scope detection (Show measurement, test conditions, and source)

Source value: 92
Score: 92
Metric: global out-of-scope detection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 natural retrieval k=5; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Higher is better
Sample: 100
Lower confidence bound: 86
Upper confidence bound: 97
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; fixed BM25 retrieval at k=5; natural candidate menu, without gold-label repair; population: public global out-of-scope test queries; exact held-out denominator N=100; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

CLINC150 natural retrieval k=10: missing-answer detection (Show measurement, test conditions, and source)

Source value: 100
Score: 100
Metric: missing-answer detection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 natural retrieval k=10; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Higher is better
Sample: 5
Lower confidence bound: 100
Upper confidence bound: 100
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; fixed BM25 retrieval at k=10; natural candidate menu, without gold-label repair; population: in-scope queries whose reference intent is absent from the menu; exact held-out denominator N=5; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

CLINC150 natural retrieval k=10: covered-input false rejection (Show measurement, test conditions, and source)

Source value: 48.8135593220339
Score: 48.8135593220339
Metric: covered-input false rejection
Unit: %
Benchmark version: arXiv:2610.03387v2; Decision_Model_Benchmark 0.2.0; commit 6e44dc7fd29c61c1ba2da89d102f4baad17fb753; CLINC150 natural retrieval k=10; Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851
Category: decision
Direction: Lower is better
Sample: 295
Lower confidence bound: 43.389830508474574
Upper confidence bound: 54.91525423728814
Source type: independent-evaluator
Evaluator: Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Laya 0.3.21 English checkpoint revision 55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851; fixed BM25 retrieval at k=10; natural candidate menu, without gold-label repair; population: reference-present test queries; exact held-out denominator N=295; 2,000-query bootstrap CI. Calibration queries are separate and excluded.
Context: These are protocol-specific candidate-selection or rejection results. Detection, false rejection, and support counts use distinct populations; they are not general deployment accuracy.

Source Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics

Page 1 of 27

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

AutoTrust JEV-27B-VL70

Clef52

GLiNER2.5-multi-Decide45

Decider 4B43

Jev43

pplx-decider-v1.1-27b42

Intern-Decision-4B40

Laya Typed Decisions40

Kev 9B40

pplx-decider-v1-27b39

Decider 0.8B39

Julia 139

Intern-Decision-0.8B38

Laya0

GLiNER2.5-multi-Decide

Decider 4B

Jev

pplx-decider-v1.1-27b

Laya

As of November 2026 · Source: gradually.ai popularity index

Over time

Weekly values from October 18, 2026 to November 6, 2026

Layad1-2B 0d1-Omni-2B 0Decider 35B A3B NVFP4 0GLiDE 0

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 compared

Show all 17 comparisons

Similar models

More models of the same kind as Laya, chosen by model line, developer, size, price and release date.

All specifications

Architecture

Model class: System One decision model; Source: Model class
Parameters: 421 million; Source: Parameters

Input and output

Input: Text, states and typed questions; Source: Input
Output: Structured probabilities, no text generation; Source: Output
Context window: 512 tokens; Source: Context window

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

Access: Local; Source: Access
License: Apache-2.0; Source: License

API

Model ID: convaiinnovations/laya; Source: Model ID

Further details

Result: Choice, Noul and Score with probabilities; Source: Result
Scope: Laya publishes separate English, multilingual and typed-decisions checkpoints.; Source: Scope
Notes: The English variant is limited outside English; the variants are local checkpoints.; Source: Notes

Sources and data date

Show all 37 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.

Type	Evidence and data date
Research status	Research date October 15, 2026. 30 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Not published	evaluated-weight-artifact-hash Note The card names the evaluated checkpoint but does not bind the comparison to an immutable weight-file hash.
Not published	evaluator-loaded-weight-digest Note The report does not disclose loaded checkpoint hash or revision. The matching public Hugging Face revision by lastModified date is only a chronology candidate.
Not published	test-dataset-revision Note Fixed samples and sample counts are reported, but dataset revision hashes are not pinned in the captured protocol.
Not published	unreported-confidence-intervals Note No CI is published for ECE or MASSIVE per-language accuracy; none was calculated or inferred.
Not published	candidate-transfer-intervals Note Table 7 publishes point rates. Its caption refers to separate uncertainty data, but the table contains no cell-level intervals and no separate CI was imported.
Not published	verdict-score-question-denominator Note The paper does not report how many of the 2,000 decisions contain a score question; MAE and W1 rows therefore have no sampleCount.
Not published	immutable-model-or-route-revision Note The zero-shot IE custom benchmark maps its exact public label to the profile task, but does not publish the loaded weight hash or immutable serving revision.
Not published	sampling-uncertainty Note The zero-shot IE small custom pools have no published confidence intervals. This gap is specific to those score cells.
Not published	Heringer Laya checkpoint revision Original wording The HTTP pilot identifies the open Laya English checkpoint but does not state the evaluated weight revision or hash.
Unresolved	Heringer bracket-range semantics Original wording Hard-tier accuracy cells print bracketed ranges, but the captured table context does not identify them as confidence intervals. They are not mapped into CI fields.
Unresolved	real-traffic and adversarial generalization Original wording The HTTP pilot explicitly uses synthetic data, one template author, and preliminary settings. It is not evidence of performance on live traffic.
Not published	Immutable Decision Index served checkpoint revision Note The current public index gives a model URL and, where available, a served_checkpoint label, but no immutable commit/digest for the exact evaluated weights.
Additional source	Laya repository (retrieved September 24, 2026) · Laya repository · Editorial description reviewed October 2, 2026
Additional source	Laya model card (retrieved September 24, 2026; October 15, 2026; October 17, 2026; October 19, 2026) · Laya model card · Editorial description reviewed October 2, 2026
Additional source	ConvAI Innovations, Laya typed-decisions model card (retrieved October 16, 2026)
Additional source	Decision_Model_Benchmark 0.2.0, raw confirmation metrics (retrieved October 16, 2026)
Additional source	Decision_Model_Benchmark 0.2.0, raw CLINC150 confirmation metrics (retrieved October 16, 2026)
Additional source	Hanno-Labs DecisionBench result repository (retrieved October 17, 2026)
Additional source	System One report, Laya AI-hosted result summary for laya, pinned source commit ed4544869a9480aa98db6572a21733803eadcbb8 (retrieved October 17, 2026)
Additional source	Monash University, candidate coverage and rejection-policy transfer paper (retrieved October 16, 2026; October 17, 2026)
Additional source	LEC AI, Verdict: Encode Once, Decide Many (retrieved October 17, 2026)
Additional source	Zero-shot IE benchmark, pinned public result snapshot (c45a00c25d2731ac7047def97b91421aaff2f611) (retrieved October 17, 2026)
Additional source	Liu, Evaluating System One Models for Agent Security Decisions, arXiv:2609.33401v1 (retrieved October 17, 2026)
Additional source	Mazzoni and Failla, Benchmarking System One Models in Online Moderation, arXiv:2610.07953v1 (retrieved October 17, 2026)
Additional source	Li, Fast Models, Slow Evidence, arXiv:2610.02267v1 (retrieved October 17, 2026)
Additional source	Laya GitHub issue #450, public result report by bhushankinge (retrieved October 17, 2026)
Additional source	sysone-bench v2.0.0 report.json (retrieved October 17, 2026)
Additional source	Duan et al., General Decision Models: Benchmarking and Insights Beyond Jev (arXiv:2610.03935v2) (retrieved October 18, 2026)
Additional source	StreamDecisionBench GitHub README (retrieved October 18, 2026)
Additional source	Niklas Heringer, System One AI for Cybersecurity: Not Every Decision Needs a Generative Model (retrieved October 18, 2026)
Additional source	Nadir, Jev, Laya, GLiNER: Beaten by len() (retrieved October 18, 2026)
Additional source	Laya Benchmarks, upstream measurement table (citing Laya BENCHMARKS.md) (retrieved October 19, 2026)
Additional source	ConvAI Innovations, Laya model card (README revision cce578c3df804585aea413f0413db0baa43dafc7) (retrieved October 21, 2026)
Additional source	local-jev-bench (retrieved October 22, 2026)
Additional source	laidick, System-One Model Benchmark (retrieved October 23, 2026)
Additional source	avbiswas, BEV Decider model card (retrieved October 23, 2026)
Additional source	Jev Decision Index public index.json (release-v4) (retrieved October 28, 2026)
Additional source	Model metadata (retrieved October 16, 2026; October 17, 2026)
Additional source	Model metadata (retrieved October 17, 2026)1 2 3 4 5
Additional source	OpenRouter: Popularity index (derived), CC BY 4.0
Additional source	Vercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional source	Ollama: cumulative downloads for the popularity index
Additional source	Replicate: cumulative runs for the popularity index
Additional source	Google Trending Now: limited feed selection for Germany and the US
Additional source	Creative Commons Attribution 4.0