DeepSWE is a new state-of-the-art open-source software engineering model trained entirely using reinforcement learning, based on Qwen3-32B. https://
together.ai/blog/deepswe Fantastic work from @togethercompute @Agentica_
RESEARCH
-

DeepSWE: New Open-Source Software Engineering Model with Reinforcement Learning
By
–
-
AI Video Models Behavior Under Research Testing Conditions
By
–
Ever wonder what AI video models do when our researchers break them? Definitely not static like an old TV screen! pic.twitter.com/Z5zXgcryfB
— Midjourney (@midjourney) 3 juillet 2025Ever wonder what AI video models do when our researchers break them? Definitely not static like an old TV screen!
-
2025: The End of Our World as We Know It
By
–
2025: The end of our world as we know it | Peter Leyden https://
youtu.be/-zoCpFfOH04?si
=1qBFoxfETRqDobe4
… via @YouTube -
Open-Source Video Audio Model Emerges as Veo 3 Alternative
By
–
We now have an open-source video + audio model like Veo 3.
— Matt Shumer (@mattshumer_) 2 juillet 2025
It’s not even close to as good, but it’s a huge step in the right direction! https://t.co/ce85FpdBdJWe now have an open-source video + audio model like Veo 3. It’s not even close to as good, but it’s a huge step in the right direction!
-
Gemini’s Vision Capabilities and State-of-the-Art Progress
By
–
A conversation with @AniBaddepudi about Gemini's vision capabilities, how we got to SOTA, and where we and the ecosystem go next. Ani is a friend and collaborator so this conversation was a lot of fun : ) pic.twitter.com/DkgX7RVOA5
— Logan Kilpatrick (@OfficialLoganK) 2 juillet 2025A conversation with @AniBaddepudi about Gemini's vision capabilities, how we got to SOTA, and where we and the ecosystem go next. Ani is a friend and collaborator so this conversation was a lot of fun : )
-
Kate Crawford praises critical artist approach to AI tools
By
–
I've long regarded @katecrawford as one of the most lucid and urgent voices in the realm of AI, so it was incredibly meaningful to hear this shoutout during her excellent conversation at @ArtBasel. To be seen by someone like Kate as a "rare" example of an artist approaching AI… pic.twitter.com/5iGPMLr5aL
— Sasha Stiles | AI alter ego Technelegy ✍️🤖 (@sashastiles) 2 juillet 2025I've long regarded @katecrawford as one of the most lucid and urgent voices in the realm of AI, so it was incredibly meaningful to hear this shoutout during her excellent conversation at @ArtBasel. To be seen by someone like Kate as a "rare" example of an artist approaching AI tools with a critical eye, asking questions, pushing boundaries in a sea of AI "spectacularism and distraction" and "cool images" – it truly means a lot.
→ View original post on X — @katecrawford, 2025-07-02 21:49 UTC
-

Math Reasoning Improvement Generalizes Across LLM Domains
By
–
benchmark-maxxing math evals can now be spotted systematically In the paper “Does Math Reasoning Improve General LLM Capabilities?” the authors show that models tuned with RL on math data can generalize their gains across domains, while SFT-tuned rarely transfer beyond math,
-

Data Quality and Curation Key to AI Model Differentiation
By
–
Models commoditize. Data differentiates.
— Snorkel AI (@SnorkelAI) 2 juillet 2025
Snorkel Expert DaaS:
👉 Find blind spots
👉 Curate data
👉 Scale HITL
👉 Control quality
Start with better data. Start with Snorkel. pic.twitter.com/P65ZYAA63wModels commoditize. Data differentiates. Snorkel Expert DaaS: Find blind spots Curate data Scale HITL Control quality Start with better data. Start with Snorkel.
-
How AI Models Learn to Process Visual Data Multimodally
By
–
How does an AI model actually learn to see? 🤖
— Google AI (@GoogleAI) 2 juillet 2025
Learn about the tech behind native multimodality, how models reason over visual data like documents and video, and the future of proactive AI assistants with @OfficialLoganK and Gemini Model Behavior Product Lead, @AniBaddepudi. ↓… pic.twitter.com/t6JmNOPHIXHow does an AI model actually learn to see? Learn about the tech behind native multimodality, how models reason over visual data like documents and video, and the future of proactive AI assistants with @OfficialLoganK and Gemini Model Behavior Product Lead, @AniBaddepudi
. ↓ -
AI Leadership Next 24 Months Shape Future Economy
By
–
The next 24 months in AI will determine who leads the next economy. And that doesn't feel like hyperbole to me.