TianyuCodingsDecision modelOpen source
NanoJev
PopularityRank 393 of 876
34out of 100-9 vs. last month
Summary
NanoJev is a 0.6-billion-parameter model from TianyuCodings that brings the interface idea of Jev into a small local game and control model. It takes a game state, a question, and candidates, then returns Choice, Boolean, or Score as direct probabilities. The computation runs in parallel without decoding answer tokens.
The released unified-games-v1 variant was trained for Maze, Snake, ViZDoom, and Predict Position, so it remains focused on those four game and control tasks. The included source code is MIT-licensed; the Qwen base retains its upstream license.
All benchmark results
Every published result with its source, test setup and comparison models.
Show all 18 published benchmark results
Page 1 of 2
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.
JEVal (Show measurement, test conditions, and source)
- Source value
- 25.34
- Score
- 25.34
- Metric
- ACC (accuracy; invalid responses count as incorrect)
- Unit
- %
- Benchmark version
- arXiv:2610.03935v2, Table 1
- Direction
- Higher is better
- Sample
- 10938
- Source type
- independent-evaluator
- Evaluator
- Duan et al., General Decision Models: Benchmarking and Insights Beyond Jev
- Status
- active
- Retrieved at
- 2026-10-23
- Methodology
- Independent evaluation. JEVal Table 1 reports 25 configurations on 10,938 choice and noul questions. The paper's implementation section says open-weight decision models run locally on NVIDIA H100 GPUs using vLLM; the Table 1 row does not identify a weight digest, reasoning effort, or NanoJev-specific decoding settings.
- Context
- Table 1 counts invalid responses as incorrect for accuracy; ECE and raw Brier apply only to valid responses. The paper does not report a confidence interval or an immutable NanoJev weight/serving revision for this row.
Held-out Maze successes (Show measurement, test conditions, and source)
- Source value
- 0.4
- Score
- 0.4
- Metric
- success rate
- Unit
- ratio
- Category
- game-decision
- Direction
- Higher is better
- Sample
- 10
- Source type
- vendor-reported
- Evaluator
- C-Tianyu NanoJev model card
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Vendor-reported result. unified-games-v1 checkpoint; matched 274-case test set; common structured observations, candidate actions and seeded epsilon-greedy controller (seed 17).
Held-out Snake successes (Show measurement, test conditions, and source)
- Source value
- 1
- Score
- 1
- Metric
- success rate
- Unit
- ratio
- Category
- game-decision
- Direction
- Higher is better
- Sample
- 8
- Source type
- vendor-reported
- Evaluator
- C-Tianyu NanoJev model card
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Vendor-reported result. unified-games-v1 checkpoint; matched 274-case test set; common observations, candidate actions and seeded controller (seed 17).
ViZDoom Basic test successes (Show measurement, test conditions, and source)
- Source value
- 1
- Score
- 1
- Metric
- success rate
- Unit
- ratio
- Category
- game-decision
- Direction
- Higher is better
- Sample
- 128
- Source type
- vendor-reported
- Evaluator
- C-Tianyu NanoJev model card
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Vendor-reported result. unified-games-v1 checkpoint; matched Basic test cases; shared observations and seeded controller.
ViZDoom Predict Position test successes (Show measurement, test conditions, and source)
- Source value
- 0.2109375
- Score
- 0.2109375
- Metric
- success rate
- Unit
- ratio
- Category
- game-decision
- Direction
- Higher is better
- Sample
- 128
- Source type
- vendor-reported
- Evaluator
- C-Tianyu NanoJev model card
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Vendor-reported result. unified-games-v1 checkpoint; matched Predict Position cases; shared observations and seeded controller.
DecisionBench all-requested primary accuracy (Show measurement, test conditions, and source)
- Source value
- 0.3577405857740586
- Score
- 0.3577405857740586
- Metric
- all-requested accuracy, unsupported and error rows count as misses
- Unit
- ratio
- Benchmark version
- Hanno DecisionBench 1.0; result commit e2d6a133cee91b9d8c6fbd076b21f32f3d75c81f; dataset b7c8107e01ecb1aee7c7eaf5caee4a3ba9f59443; task spec a477586be2d19dc109c9f2479e3c455b76e6311758c05f42b5a0ca15f4c34b44; C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax
- Category
- decision
- Direction
- Higher is better
- Sample
- 23900
- Source type
- independent-evaluator
- Evaluator
- Hanno-Labs DecisionBench result repository
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; 600000000 parameters; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax; Base Qwen/Qwen3-0.6B with structured decision heads; the Hanno submission does not name a specialized repository subdirectory, so no game or navigation variant is inferred. Submitted 2026-09-20T15:09:15Z.
- Context
- This is one vendor-submitted run on the evaluator's benchmark, not a multi-seed estimate. No absolute-score confidence interval is reported. The result binds the repository revision but not a specialized subdirectory; no variant-specific evidence is inferred.
DecisionBench supported-output accuracy (Show measurement, test conditions, and source)
- Source value
- 0.37151299209177024
- Score
- 0.37151299209177024
- Metric
- accuracy on successful supported outputs
- Unit
- ratio
- Benchmark version
- Hanno DecisionBench 1.0; result commit e2d6a133cee91b9d8c6fbd076b21f32f3d75c81f; dataset b7c8107e01ecb1aee7c7eaf5caee4a3ba9f59443; task spec a477586be2d19dc109c9f2479e3c455b76e6311758c05f42b5a0ca15f4c34b44; C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax
- Category
- decision
- Direction
- Higher is better
- Sample
- 23014
- Source type
- independent-evaluator
- Evaluator
- Hanno-Labs DecisionBench result repository
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; 600000000 parameters; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax; Base Qwen/Qwen3-0.6B with structured decision heads; the Hanno submission does not name a specialized repository subdirectory, so no game or navigation variant is inferred. Submitted 2026-09-20T15:09:15Z.
- Context
- This is one vendor-submitted run on the evaluator's benchmark, not a multi-seed estimate. No absolute-score confidence interval is reported. The result binds the repository revision but not a specialized subdirectory; no variant-specific evidence is inferred.
DecisionBench output coverage (Show measurement, test conditions, and source)
- Source value
- 0.962928870292887
- Score
- 0.962928870292887
- Metric
- successful supported rows divided by requested rows
- Unit
- ratio
- Benchmark version
- Hanno DecisionBench 1.0; result commit e2d6a133cee91b9d8c6fbd076b21f32f3d75c81f; dataset b7c8107e01ecb1aee7c7eaf5caee4a3ba9f59443; task spec a477586be2d19dc109c9f2479e3c455b76e6311758c05f42b5a0ca15f4c34b44; C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax
- Category
- decision
- Direction
- Higher is better
- Sample
- 23900
- Source type
- independent-evaluator
- Evaluator
- Hanno-Labs DecisionBench result repository
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; 600000000 parameters; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax; Base Qwen/Qwen3-0.6B with structured decision heads; the Hanno submission does not name a specialized repository subdirectory, so no game or navigation variant is inferred. Submitted 2026-09-20T15:09:15Z.
- Context
- This is one vendor-submitted run on the evaluator's benchmark, not a multi-seed estimate. No absolute-score confidence interval is reported. The result binds the repository revision but not a specialized subdirectory; no variant-specific evidence is inferred.
DecisionBench expected calibration error (Show measurement, test conditions, and source)
- Source value
- 0.15484657565150664
- Score
- 0.15484657565150664
- Metric
- 15-bin top-label expected calibration error
- Unit
- ratio
- Benchmark version
- Hanno DecisionBench 1.0; result commit e2d6a133cee91b9d8c6fbd076b21f32f3d75c81f; dataset b7c8107e01ecb1aee7c7eaf5caee4a3ba9f59443; task spec a477586be2d19dc109c9f2479e3c455b76e6311758c05f42b5a0ca15f4c34b44; C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax
- Category
- decision
- Direction
- Lower is better
- Source type
- independent-evaluator
- Evaluator
- Hanno-Labs DecisionBench result repository
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; 600000000 parameters; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax; Base Qwen/Qwen3-0.6B with structured decision heads; the Hanno submission does not name a specialized repository subdirectory, so no game or navigation variant is inferred. Submitted 2026-09-20T15:09:15Z.
- Context
- The result JSON does not state a metric-specific N for ECE; none is inferred. One run and no confidence interval are reported.
DecisionBench mean negative log likelihood (Show measurement, test conditions, and source)
- Source value
- 1.942477124266686
- Score
- 1.942477124266686
- Metric
- mean negative log likelihood
- Unit
- No unit provided
- Benchmark version
- Hanno DecisionBench 1.0; result commit e2d6a133cee91b9d8c6fbd076b21f32f3d75c81f; dataset b7c8107e01ecb1aee7c7eaf5caee4a3ba9f59443; task spec a477586be2d19dc109c9f2479e3c455b76e6311758c05f42b5a0ca15f4c34b44; C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax
- Category
- decision
- Direction
- Lower is better
- Source type
- independent-evaluator
- Evaluator
- Hanno-Labs DecisionBench result repository
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; 600000000 parameters; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax; Base Qwen/Qwen3-0.6B with structured decision heads; the Hanno submission does not name a specialized repository subdirectory, so no game or navigation variant is inferred. Submitted 2026-09-20T15:09:15Z.
- Context
- The result JSON does not state the log base or metric-specific N; neither is inferred. One run and no confidence interval are reported.
DecisionBench English suite: supported accuracy (Show measurement, test conditions, and source)
- Source value
- 0.3725578464031692
- Score
- 0.3725578464031692
- Metric
- accuracy
- Unit
- ratio
- Benchmark version
- Hanno DecisionBench 1.0; result commit e2d6a133cee91b9d8c6fbd076b21f32f3d75c81f; dataset b7c8107e01ecb1aee7c7eaf5caee4a3ba9f59443; task spec a477586be2d19dc109c9f2479e3c455b76e6311758c05f42b5a0ca15f4c34b44; C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax
- Category
- decision
- Direction
- Higher is better
- Sample
- 22214
- Source type
- independent-evaluator
- Evaluator
- Hanno-Labs DecisionBench result repository
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; 600000000 parameters; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax; Base Qwen/Qwen3-0.6B with structured decision heads; the Hanno submission does not name a specialized repository subdirectory, so no game or navigation variant is inferred. Submitted 2026-09-20T15:09:15Z. English suite view DecisionBench(eng, v1); supported-output denominator 22214.
- Context
- This is an English-only view inside the same run, not an independent run. No confidence interval is reported.
DecisionBench English suite: expected calibration error (Show measurement, test conditions, and source)
- Source value
- 0.15798160879377027
- Score
- 0.15798160879377027
- Metric
- expected calibration error
- Unit
- ratio
- Benchmark version
- Hanno DecisionBench 1.0; result commit e2d6a133cee91b9d8c6fbd076b21f32f3d75c81f; dataset b7c8107e01ecb1aee7c7eaf5caee4a3ba9f59443; task spec a477586be2d19dc109c9f2479e3c455b76e6311758c05f42b5a0ca15f4c34b44; C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax
- Category
- decision
- Direction
- Lower is better
- Sample
- 22214
- Source type
- independent-evaluator
- Evaluator
- Hanno-Labs DecisionBench result repository
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. C-Tianyu/NanoJev@4a19595eada0857133c0d2be024f879a4077054b; 600000000 parameters; adapter nanojev-hf; probability source native_candidate_logits_fp32_softmax; Base Qwen/Qwen3-0.6B with structured decision heads; the Hanno submission does not name a specialized repository subdirectory, so no game or navigation variant is inferred. Submitted 2026-09-20T15:09:15Z. English suite view DecisionBench(eng, v1); supported-output denominator 22214.
- Context
- This is an English-only view inside the same run, not an independent run. No confidence interval is reported.
Page 1 of 2
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
NanoJevDecider 2B Vision 36Jeff 1 36GLiNER2.5-Decide-1B 34Decider 35B A3B 34
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.
Similar models
More models of the same kind as NanoJev, chosen by model line, developer, size, price and release date.
- Decider 0.8BMapika · similar size
- Intern-Decision-0.8BInternLM · similar size
- Laya Typed DecisionsConvai Innovations · similar size
- GLiNER2.5-DecideFastino Labs · similar size
- GLiNER2.5-Decide-1BFastino Labs · similar size
- Laya MultilingualConvai Innovations · similar size
- LayaConvai Innovations · similar size
- GLiNER2.5-multi-DecideFastino Labs · 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 released variant is trained on the documented game tasks and is not a general classifier.
- Source: Notes
Sources and data date
Show all 18 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.