# Nemotron 3 Ultra ships, MiniMax M3, Gemma 4 12B, Glasswing grows

> NVIDIA releases a 550B open agent model at Computex, MiniMax and Google add open models, and Anthropic widens Project Glasswing.

Published 13 Oct 2026. Coverage: 1 Jun 2026 to 7 Jun 2026. By Ali Asif.

## Items

1. **Nemotron 3 Ultra** (NVIDIA, Open weights): A 550B mixture-of-experts with 55B active, a 1M-token context and NVFP4 pretraining, released June 4 with NVIDIA claiming 5.9 times the throughput of GLM-5.1. [Primary source](https://research.nvidia.com/labs/nemotron/Nemotron-3-Ultra/)
2. **MiniMax M3** (MiniMax, Open weights): An open model with a 1M-token context built on a sparse attention scheme, listed around June 1. [Primary source](https://huggingface.co/MiniMaxAI/MiniMax-M3)
3. **Gemma 4 12B** (Google, Open weights): A June 3 mid-size Gemma that takes image and audio without separate encoders and runs on a 16GB laptop, under Apache 2.0. [Primary source](https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemma-4-12B/)
4. **Expanding Project Glasswing** (Anthropic, Frontier): On June 2 Anthropic added about 150 organizations in more than 15 countries and reported over 10,000 high or critical flaws found by partners. [Primary source](https://www.anthropic.com/news/expanding-project-glasswing)
5. **GTC Taipei at Computex 2026** (NVIDIA, Robotics): GR00T 1.7 in early access, Isaac Sim 6.0 generally available and the first open humanoid reference design. [Primary source](https://blogs.nvidia.com/blog/nvidia-gtc-taipei-computex-2026-news/)
6. **Sites preview in the Codex app** (OpenAI, Agents and coding): June 2 brought a preview for building and deploying hosted websites and web apps from Codex. [Primary source](https://developers.openai.com/codex/sites)
7. **Cosmos 3: Omnimodal World Models for Physical AI** (arXiv, Image, video, world): The technical report behind NVIDIA's new world model. [Primary source](https://arxiv.org/abs/2606.02800)
8. **Qwen-VLA** (arXiv, Robotics): A vision-language-action model meant to cover many tasks, environments and robot bodies. [Primary source](https://arxiv.org/abs/2605.30280)

Computex week, so NVIDIA set the pace, with Anthropic and Google both landing news around it.

## Nemotron 3 Ultra

NVIDIA announced Nemotron 3 Ultra at Sunday night's keynote and released it on Thursday, June 4. It is a 550B mixture-of-experts model with 55B active parameters, a hybrid Mamba-attention design, a context of up to 1M tokens, and NVFP4 pretraining. Checkpoints in NVFP4, BF16 and base form are out, with a reward-model checkpoint for RLHF. NVIDIA reports throughput 5.9 times GLM-5.1, 4.8 times Kimi K2.6 and 1.6 times Qwen 3.5 397B on long generations, and accuracy on par with other leading open models. All of that is NVIDIA's own testing. Early adopters it names include Perplexity, Palantir, ServiceNow, CrowdStrike and Harvey.

The pitch is cost per agent step. If throughput holds up in other people's hands, open models at this size become a fair option for the planning layer of an agent, and the cheap workers stop being the only job for them.

## Gemma 4 12B and MiniMax M3

Google released Gemma 4 12B on June 3 as an Apache 2.0 model that takes text, image and audio in one architecture with no separate encoders, small enough for 16GB of memory. MiniMax M3 listed around June 1 with a million-token context and a sparse attention design; I have not seen its weights tested yet.

## Glasswing, one more time

On June 2 Anthropic widened Project Glasswing by about 150 organizations across more than 15 countries, from an initial cohort of roughly 50. Partners have found over 10,000 high or critical flaws, it says. New commitments include Claude Security, a product on Opus 4.8 that scans codebases and suggests patches, and a larger Cyber Verification Program. Priority goes to essential infrastructure providers and open-source maintainers.

## Robots and tools

Computex also brought GR00T 1.7 in early access, Isaac Sim 6.0 generally available, and an open humanoid reference design on Jetson Thor. OpenAI added a Sites preview to Codex on June 2. The Cosmos 3 technical report and the Qwen-VLA paper filled the research side.

What I'd watch next: Google said Gemini 3.5 Pro would arrive this month, so that is the date I have circled.

## My take

Open weights keep creeping up the size ladder, and the part I care about is the 12B Gemma that runs in 16GB. I can put that on a laptop in a site office with no connection back to anyone. Glasswing growing to about 150 more organizations also reads like the cyber side is settling into a real program, not a one-off demo.

## Sources

- https://research.nvidia.com/labs/nemotron/Nemotron-3-Ultra/
- https://huggingface.co/MiniMaxAI/MiniMax-M3
- https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemma-4-12B/
- https://www.anthropic.com/news/expanding-project-glasswing
- https://blogs.nvidia.com/blog/nvidia-gtc-taipei-computex-2026-news/
- https://developers.openai.com/codex/sites
- https://arxiv.org/abs/2606.02800
- https://arxiv.org/abs/2605.30280

Canonical: https://heyaliasif.com/ai/2026-06-01-week-in-ai