2/ What changed: old Cosmos split the work across separate models — one to understand a scene, one to generate video, one for controlled simulation.
Cosmos 3 fuses everything into a single Mixture-of-Transformers with two towers:
→ a reasoner (the VLM "brain")
→ a diffusion
@kimmonismus
-
Cosmos 3 merges reasoner and diffusion towers in one model
By
–
-
NVIDIA open-sources Cosmos 3, first open omnimodel for physical AI
By
–
1/ NVIDIA just open-sourced Cosmos 3 at GTC Taipei!
— Chubby♨️ (@kimmonismus) 1 juin 2026
It's the first fully open "omnimodel" for physical AI – one model that understands the real world, predicts what happens next, and generates the actions a robot should take.
Weights, code, datasets. All open. And this is… https://t.co/5Y8BbXUWqJ pic.twitter.com/3bOMlwO0B21/ NVIDIA just open-sourced Cosmos 3 at GTC Taipei! It's the first fully open "omnimodel" for physical AI – one model that understands the real world, predicts what happens next, and generates the actions a robot should take. Weights, code, datasets. All open. And this is
-
User discovers how to re-enable the context window circle in Codex
By
–
lol just figured out you can re-enable the context window circle in codex. thank god
-
Claude Mythos pricing and upcoming model cost prediction
By
–
Claude Mythos is $25 per million input tokens and $125 per million output tokens. I assume that the Mythos-like model that Anthropic will release in the coming weeks will be just as expensive. lets see
-
Apple AI to run distilled Google Gemini on iPhone
By
–
Interesting updates on Apple AI: As Apple's WWDC lands next month, and the long-delayed Siri and on-device AI upgrades are expected to be the centerpiece: a smaller, distilled version of Google's Gemini running locally on iPhone silicon, pitched on privacy and lower token costs.
-
AI in cars: Tesla integrates Grok inspired by Knight Rider
By
–
What is going to be a real game changer is AI in cars. Tesla is leading the way; the integration of Grok into the Tesla OS enables seamless experiences. When I was a child, I loved *Knight Rider*, and the idea that I might someday be able to talk to my car seemed like a wild
-
Opus 4.8 improves, but GPT-5.5 xhigh beats it for cheaper cost
By
–
Opus 4.8 is a solid jump over Opus 4.7 on DeepSWE, while also lowering the average cost per task.
— Chubby♨️ (@kimmonismus) 31 mai 2026
However, GPT-5.5 xhigh still beats it by a pretty clear margin while being cheaper.
OpenAI has been cooking insanely hard with its models lately. Really excited to see what GPT-5.6… https://t.co/UC7Rl2cX6lOpus 4.8 is a solid jump over Opus 4.7 on DeepSWE, while also lowering the average cost per task. However, GPT-5.5 xhigh still beats it by a pretty clear margin while being cheaper. OpenAI has been cooking insanely hard with its models lately. Really excited to see what GPT-5.6
-
Microsoft Builds Super App to Unify Fragmented Copilots
By
–
Fortune reports Microsoft is building a "super app" to unify its scattered Copilots. Under 4.5% of 450 million Microsoft 365 seats pay for Copilot. Around 20 million, out of nearly half a billion. The app is pitched as a fix for fragmentation. The open question is whether
-
Seedance 2.0 still unbeaten in text-to-video since February
By
–
I still find it crazy that no lab has surpassed Seedance 2.0 in text-to-video, even though Seedance 2.0 was released back in February.
-
GPT-5.6 improvement and token efficiency in agentic workflows
By
–
It’s reasonable to expect that the next iteration will be better. It would be surprising if GPT-5.6 wasnt an improvement over GPT-5.5. But the more interesting part is token efficiency. As models move into more complex, longer-running, agentic workflows, every wasted token