On 4 August 2026, NVIDIA announced Alpamayo 2 Super. The model combined trajectory generation, chain-of-causation traces, meta-actions, auto-labelling and grounded visual question answering.
This release brief was checked against first-party material on 10 August 2026. The date above is the public announcement date, not the date a repository was created or a third-party provider added the model. Where access or weights arrived later, that distinction is recorded below.
Release record
| Field | Verified detail |
|---|---|
| Announcement | 4 August 2026 |
| Availability or weight release | 4 August 2026 as commercially usable weights on Hugging Face |
| Release type | open-weight multimodal autonomous-driving foundation model |
| Access | downloadable under OpenMDW 1.1 with commercial fine-tuning and redistribution rights |
| Architecture | Cosmos 3 Super Reasoner-based multimodal model with full-surround camera fusion and reinforcement post-training |
| Maximum stated context | full-surround driving clips; token context not stated as the launch headline |
What changed
Alpamayo 2 Super moves an open physical-AI model from research-only use toward commercial deployment. Its five coupled outputs make driving decisions more inspectable, and its scale supports teacher-data generation and distillation into in-vehicle models.
The practical comparison is therefore not simply whether Alpamayo 2 Super has the largest headline score. Teams need to ask whether its architecture, access terms, latency, tool behaviour and evaluation setup match the workload they actually intend to run. A model can lead one harness while losing on cost, refusal behaviour, multilingual quality or repeatability in another.
Benchmarks worth retaining
| Evaluation | Reported result | How to read it |
|---|---|---|
| LingoQA / Lingo-Judge | ranked first among nearly 40 evaluated models | NVIDIA-reported autonomous-driving reasoning comparison |
| Qwen2.5-VL 72B margin | +17.0 points | First-party Lingo-Judge comparison |
| Model scale | 3x the 10B-parameter Alpamayo 1.5 and 1 models | Relative size disclosure, not a safety metric |
These are release-time results, not independently reproduced guarantees. NVIDIA’s benchmark comparisons use its disclosed driving harnesses and do not constitute a safety certification or production autonomy approval. Scores should remain attached to the disclosed effort setting, agent harness, tool access, timeout, context-management policy and judge model. Moving a number into a procurement sheet without those conditions creates false comparability.
Architecture and access
Alpamayo 2 Super is described as Cosmos 3 Super Reasoner-based multimodal model with full-surround camera fusion and reinforcement post-training with full-surround driving clips; token context not stated as the launch headline of stated context. Its access position at verification time is downloadable under OpenMDW 1.1 with commercial fine-tuning and redistribution rights. That wording matters: open weights, source-available weights, an API, a product preview and a research demonstration give adopters very different rights and different levels of reproducibility.
Before deployment, record the exact model identifier or checkpoint, inference stack, quantisation, reasoning setting, region, price schedule and supplier terms. If the release uses a custom licence, read the licence itself rather than relying on the word “open” in launch copy. If it is API-only, preserve the dated documentation and change-notice route because the served snapshot can change without a downloadable artefact.
What an evaluation should test next
For Alpamayo 2 Super, a credible internal gate should include:
- a frozen set of representative tasks with pass, fail and abstain criteria;
- a matched baseline using the same tools, timeout, prompt budget and reviewer rubric;
- repeated runs to expose variance rather than reporting a single best attempt;
- latency, token use and total task cost alongside task success;
- adversarial, multilingual and long-context cases relevant to the real deployment; and
- rollback evidence showing the previous model can be restored safely.
The wider model change-control guide explains how to keep model, prompt, tool and corpus changes reconstructable. The AI dependency inventory guide covers the release and supplier records needed after deployment.
AIEngine verdict
This is a major open specialist model and should be in the archive. AV teams must treat it as a development foundation inside a larger validation and safety case.
This is a launch assessment, not a certification. Benchmark leadership is useful evidence of where to test; it is not authorization to place the model in a high-impact workflow without domain evaluation, security review and an accountable owner.
Primary sources
Image provenance
Hero image: NVIDIA official release artwork. The locally served WebP is a crop of the first-party release or model-card asset recorded in the repository provenance manifest.
