Pin source and exact files
A model name is not an immutable checkpoint.
Record SHA, original files, tokenizer and processor.
Upstream updates change behavior.
NVIDIA · FIELD VERIFIED
Primary sources last checked on 16 Nov 2026. Verification freshness is tracked separately from factual change, and unknown values remain explicit rather than inferred.
MODEL INTELLIGENCE · SOURCE-FIRST
Open Model Weights records Cosmos3-Nano-Sim-Auto as a 15.157B-parameter action-conditioned driving sensor simulation open-weight model in the Cosmos 3 Nano family, with NVIDIA listed as the developer. The checked source repository provides 10 recognized weight artifacts, format evidence for Safetensors and a context window that is not declared in the checked standard sources. Values that are not evidenced in the checked sources remain explicitly undisclosed or unknown.
The repository declares OpenMDW 1.1. Commercial use requires compliance with publisher-specific terms. Open Model Weights reports the checked license evidence and classification; this is not legal advice.
No base model is declared in the checked standard lineage metadata, so Open Model Weights does not infer one. Training recipe status: not-verified; data disclosure status: not-verified.
+ Save to my shortlistMonitor changes →Local browser monitoring · no background email alerts
DIRECT ANSWER · FEATURED MODEL BRIEFING
Source-grounded editorial layer · reviewed 2026-11-16
NVIDIA's causal-distilled Cosmos 3 Nano streaming driving simulator with action-conditioned future observations. This repository publicly lists the original Safetensors files at its pinned source revision. Model-specific permissions and attribution or redistribution conditions apply. No independent OMW benchmark has been performed.
WHY IT MATTERS
NVIDIA's causal-distilled Cosmos 3 Nano streaming driving simulator with action-conditioned future observations.
This repository publicly lists the original Safetensors files at its pinned source revision.
Model-specific permissions and attribution or redistribution conditions apply.
MODEL-SELECTION VIEW
NVIDIA's causal-distilled Cosmos 3 Nano streaming driving simulator with action-conditioned future observations. This exact checkpoint is distinguished from a general family release and from any externally hosted inference endpoint.
The developer's task framing is action-conditioned driving sensor simulation. Open Model Weights has not executed this checkpoint or independently reproduced publisher-supplied performance figures.
The official Safetensors metadata describes 15.157 billion stored tensor parameters. The publisher separately labels the logical model 16B total; this is a different accounting basis.
No verified operational context length was established from the accessible source configuration; any publisher context figure must remain attributed.
At immutable revision bea7452062866fdd8eebe4c38f305d02baed3699, the official repository lists 10 source weight or index paths. This repository publicly lists the original Safetensors files at its pinned source revision.
File names, the exact SHA and auxiliary configuration identify a reproducible source checkpoint; they do not demonstrate that the model loads in a specific inference engine or fits a given GPU memory budget.
This checkpoint declares OpenMDW 1.1. Model-specific permissions and attribution or redistribution conditions apply. The original publisher license at https://openmdw.ai/license/1-1/ governs actual use, distribution and derivative works.
Separately distributed software, datasets, inference services and outputs may raise additional questions. Operators must examine notices, permissions and any jurisdiction-specific deployment rules rather than assuming a blanket Apache license.
NVIDIA's causal-distilled Cosmos 3 Nano streaming driving simulator with action-conditioned future observations. None of the publisher benchmark charts or launch demonstrations should be represented as independent Open Model Weights measurements of these exact files.
A reproducible quality comparison must fix task datasets, prompt and input preprocessing, decoding controls, version-pinned weights, hardware and serving backend, and it must report unsuccessful cases. APIs and public weight checkpoints may not be evaluated identically.
This repository publicly lists the original Safetensors files at its pinned source revision. Before implementation, choose a compatible framework and verify the checked revision, required processor and tokenizer files, and OpenMDW 1.1 obligations.
Run representative task tests with actual peak memory, latency, correctness and failure logs. Weight-only storage arithmetic excludes activations, KV cache, sharding, batch size, runtime overhead and quantized checkpoint details.
IMPLEMENTATION & EVALUATION
Reproducible source review and operator checks; this is not a benchmark.
A model name is not an immutable checkpoint.
Record SHA, original files, tokenizer and processor.
Upstream updates change behavior.
Model rights depend on exact published terms.
Check the original publisher license.
Unlawful redistribution or use.
Parameters do not establish runtime memory.
Measure model load, memory and response quality.
Unsupported backends and OOM failures.
Vendor comparisons are not independent.
Run held-out tasks and record settings, failures and baselines.
Unsupported rankings.
Proposed tests, not OMW benchmark results. Read the official model card at the pinned source revision ↗ Inspect OMW field evidence ↗
PRIMARY & INDEPENDENT SOURCES
FAQ
NVIDIA's causal-distilled Cosmos 3 Nano streaming driving simulator with action-conditioned future observations.
This repository publicly lists the original Safetensors files at its pinned source revision.
Model-specific permissions and attribution or redistribution conditions apply.
No accessible configuration-backed context figure is verified.
OMW has not run an independent checkpoint benchmark; publisher results remain attributed.
Use official repository https://huggingface.co/nvidia/Cosmos3-Nano-Sim-Auto and exact immutable revision bea7452062866fdd8eebe4c38f305d02baed3699, including model-specific license and filenames.
03 · PRACTICAL TECHNICAL PROFILE
Inspect weight-memory estimates, model lineage and runtime evidence separately from promotional or untested performance claims.
DEPLOYMENT PROFILE
These bars visualize weight-only decimal-GB estimates. They exclude KV cache, activations, runtime overhead, optimizer state and sharding.
LICENSE & USE
Commercial use requires compliance with publisher-specific terms
This section summarizes the checked repository evidence. It is not legal advice. Inspect license evidence ↗
LINEAGE & PROVENANCE
Open Model Weights does not infer a hidden base model or training corpus when the checked sources do not declare one.
RUNTIME SIGNALS
Declared and model-card-mentioned runtimes are discovery signals. They are not treated as Open Model Weights runtime tests unless explicitly measured.
EVIDENCE TRAIL
This is where the article remains auditable: source links on one side, the registry's observed change history on the other.
Official repository SHA, weight artifact names and original model license checked.
04 · EVIDENCE AND UPDATES
Inspect dated source checks and any differences between published claims and the evidence; these checks are not independent model benchmarks.
SOURCE-BASED VISUAL BRIEF · NOTABLE
Three evidence-linked visual summaries: parameter scale, illustrative precision memory and verifiable provenance. No performance ranking is implied.
Active parameter count not established.
Repository-derived or publisher-described parameter countFRESHNESS & VERIFICATION
A verification check does not pretend the model changed. Open Model Weights updates the verification date after a real primary-source check, while factual-update dates move only when retained source evidence changes.
CONFLICT INTELLIGENCE
05 · SOURCE RECORD
Full repository-level attributes, licensing, format details and provenance are available below without hiding or rewriting the original data.
01 · WEIGHTS
lidar_vae/diffusion_pytorch_model.safetensorsmodel.safetensors.index.jsontransformer/diffusion_pytorch_model-00001-of-00006.safetensorstransformer/diffusion_pytorch_model-00002-of-00006.safetensorstransformer/diffusion_pytorch_model-00003-of-00006.safetensorstransformer/diffusion_pytorch_model-00004-of-00006.safetensorstransformer/diffusion_pytorch_model-00005-of-00006.safetensorstransformer/diffusion_pytorch_model-00006-of-00006.safetensorstransformer/diffusion_pytorch_model.safetensors.index.jsonvae/diffusion_pytorch_model.safetensors02 · LICENSE
Classification describes the checked license terms; it is not legal advice.
License / model-card evidence ↗03 · HARDWARE
Weight-only decimal-GB arithmetic, not measured peak GPU memory or file size.
04 · FORMATS
“Not observed” means no matching repository artifact/name/dtype signal was found; it does not prove no third-party conversion exists.
05 · LINEAGE
06 · TRAINING ASSETS
07 · RUNTIME SUPPORT
Open Model Weights separates model-repository evidence from external runtime-release evidence. An external support statement does not change the official model repository's weight formats and does not imply an Open Model Weights inference benchmark.
08 · SOURCES
09 · VERIFICATION HISTORY
Official repository SHA, weight artifact names and original model license checked.
MODELS FOR AI AGENTS · ROLE EVIDENCE
Specialized model. A supporting component. This checkpoint does not, by category alone, serve as the planning and tool-execution controller.
The curated notable tier is not an agent-readiness rating. Model-specific tool-call success, long-horizon task accuracy and compatible runtime configurations are not independently established here.
Source-linked license: OpenMDW 1.1 (allowed-with-conditions commercial-use classification; see original terms).
Context: Unknown or not declared.
BF16 weights: 30.31 estimated decimal GB, not runtime VRAM.
Compare agent model candidates → · Agent evaluation method → · Inspect original model repository ↗
These records are separately sourced. They do not replace independent model-plus-runtime tests.
No model-specific agent capability evaluation is recorded for this checkpoint. Missing evidence is unknown, not a failed task.
Read evidence details and benchmarks → · Use the agent model selector →