- Best-in-class reasoning and writing
- Strong ecosystem and integrations
- Advanced multimodal capabilities
OpenAI has launched GPT-6 Astra, its most capable artificial intelligence model to date. The model can operate computers, complete multistep workflows and handle tasks that would otherwise require continuous human input.
Google DeepMind equips its Gemini 3 Flash model with a new capability called “Agentic Vision.” The model is designed to actively investigate images rather than merely observe them passively — although the feature does not yet work fully automatically in all cases.
The AI research institute Allen AI (Ai2) has released SERA, a family of open-source coding agents designed to be easily adapted to private codebases at low cost. The flagship model, SERA-32B, solves up to 54.2% of tasks on the SWE-Bench-Test Verified coding benchmark (with 64K context), outperforming comparable open-source models.
There Is a New Best Math Model. OpenAI’s GPT-5.2 Pro Sets a New Record on FrontierMath
Google DeepMind’s new AI model, D4RT, is designed to reconstruct dynamic scenes from videos in four dimensions, outperforming previous methods by a wide margin.
Google DeepMind Accelerates Startup Acquisitions in AI Push
OpenAI plans to release several updates related to its Codex code model next month — and the announcement comes with a warning
Baidu Launches Ernie 5.0 — A Multimodal AI Model Ranking Among the World’s Top 10
Meta’s new AI team, Meta Superintelligence Labs, completed its first major AI models internally this month. This was stated by Chief Technology Officer Andrew Bosworth on the sidelines of the World Economic Forum in Davos in comments to Reuters. According to Bosworth, the models are “very good,” despite the team having worked for only about six months. He did not disclose technical details.
A new study reveals a fundamental weakness in current AI systems: even the most powerful multimodal language models fail at basic visual tasks that toddlers solve effortlessly.