According to NVIDIA's official blog, Alpamayo 2 Super has now entered commercial use as part of the Alpamayo series, which NVIDIA describes as the most widely adopted open autonomous-driving reasoning model family on Hugging Face. The model is distributed under OpenMDW-1.1, a permissive license issued by the Linux Foundation for open AI model releases, and NVIDIA states this license covers fine-tuning, derivative models, and commercial re-release.
ITHome's coverage adds the licensing timeline and scope: OpenMDW was first introduced in 2025 by the Linux Foundation together with the PyTorch Foundation, and the updated OpenMDW-1.1 was released by the Linux Foundation in May 2026. Per ITHome, NVIDIA has now extended OpenMDW-1.1 to cover the entire Alpamayo model family, meaning self-driving developers, automakers, and component suppliers can fine-tune the models with their own data, build derivative versions, and use modified models in commercial products.
ITHome further reports that users of Alpamayo models can freely use, modify, fine-tune, and redistribute the materials — including for commercial purposes — without applying to NVIDIA for separate permission. The one condition: anyone redistributing original or modified model materials must include the license documentation and retain the applicable original copyright and source notices.
NVIDIA's blog states that Alpamayo 2 Super is built on NVIDIA Cosmos 3 Super Reasoner and was post-trained using reinforcement learning. On scale, NVIDIA reports that Alpamayo 2 Super is three times the size of its two predecessors — Alpamayo 1.5 and Alpamayo 1 — which each carry 10 billion parameters.
Per NVIDIA's blog, Alpamayo 2 Super produces five tightly linked outputs for every driving scenario: a trajectory describing the vehicle's planned path; a chain-of-causation (CoC) trajectory that explains the reasoning behind a decision; a Meta-Action capturing the model's intent, such as yielding, changing lanes, or parking; reasoning auto-labels that generate chain-of-causation annotations for training and validation data; and visual question-answering responses combined with 2D visual grounding.
NVIDIA states that the chain-of-causation trajectory specifically integrates with the NVIDIA Halos safety verification workflow and supports AI safety practices aligned with the ISO/PAS 8800 standard — directly tying the model's reasoning output to its safety-compliance pathway.
On the LingoQA autonomous-driving reasoning benchmark, NVIDIA reports that Alpamayo 2 Super ranked first among roughly 40 evaluated models. Using its own Lingo-Judge metric, NVIDIA states the model's point-gap lead over three named competitors as follows:
| Competing model | Alpamayo 2 Super's lead (Lingo-Judge points) |
|---|---|
| Qwen2.5-VL 72B | 17.0 |
| Gemini 2.5 Pro | 15.1 |
| GPT-4o | 23.2 |
Separately, NVIDIA's blog reports that downloads of the Alpamayo series on Hugging Face have surpassed 500,000, which NVIDIA cites as reinforcing the series' position as the most widely used open autonomous-driving reasoning model family on the platform.
NVIDIA's blog lists a supporting toolset released alongside the model family: NVIDIA AlpaSim for closed-loop simulation, NVIDIA AlpaGym for high-throughput reinforcement learning, NVIDIA's open Physical AI datasets for training and testing, and open training recipes plus automated labeling pipelines intended to speed up model development, training, and validation.
That tooling matters against a caveat ITHome reports directly: Alpamayo 2 Super itself is primarily built for cloud-based development and is not a complete driving system that can be installed directly into a self-driving car. Before deployment in an actual vehicle, ITHome notes the model still requires adaptation and safety validation tailored to a specific vehicle model, sensor configuration, and road environment.
The evidence points to two tracks moving at different speeds. On one track, NVIDIA has widened commercial access — OpenMDW-1.1 now spans the whole Alpamayo family, redistribution requires no NVIDIA approval per ITHome, and the series has already logged over 500,000 Hugging Face downloads alongside a first-place LingoQA ranking among roughly 40 models. On the other track, ITHome's reporting is explicit that the released model is a cloud-development artifact, not a road-ready driving system: reaching an actual vehicle still requires vehicle-specific, sensor-specific, and road-specific adaptation and safety validation. NVIDIA's own safety framing — the chain-of-causation trajectory's integration with the Halos workflow and ISO/PAS 8800 — addresses part of that gap on paper, but the AlpaSim, AlpaGym, and open dataset tooling NVIDIA ships alongside the model are, per the evidence, exactly the kind of infrastructure that validation step would draw on.
No. ITHome reports that users can freely use, modify, fine-tune, and redistribute Alpamayo materials for commercial purposes without applying to NVIDIA for permission, though redistribution requires including the license documentation and retaining the original copyright and source notices.
No. ITHome reports that Alpamayo 2 Super is primarily built for cloud-based development rather than as a complete driving system, and that it requires adaptation and safety validation for specific vehicle models, sensor configurations, and road environments before real-world deployment.
According to Intel's newsroom announcement, the company is investing €5 billion ($5.7 billion) to expand its Leixlip campus in Ireland, producing Intel Xeon 6 and next-generation Xeon processors on the Intel 3 node. Intel says the capital programme began earlier this year and builds on more than €30 billion invested in Ireland since 1989.
According to TechCrunch and VentureBeat, Meta released Muse Code — a beta terminal-based AI coding agent for large repositories — alongside the Muse Spark 1.2 model on August 5, 2026. Muse Code installs with a single command, runs persistent background agents, and is entirely proprietary, putting Meta in direct competition with Anthropic's Claude Code and OpenAI's Codex.
Asian equities sold off again on August 6, according to Liberty Times (自由時報): South Korea's KOSPI opened down 1.81% and later retested a low of 6,379.35, off 3.31%, while Japan's Nikkei 225 broke below the 66,000 level, falling as much as 1.64% to 65,208. The declines followed a mixed Wall Street close on August 5, when the Philadelphia Semiconductor Index (SOX) dropped 1.4%. According to Anue (鉅亨網), a similar episode occurred on June 14, when Trump's move to resume an Iran blockade drove oil prices up nearly 10% and briefly sent KOSPI down close to 3% and the Nikkei down more than 1,000 points before both indexes rebounded.