Global AI Native Industry Insights – 20260521 – OpenAI | Anthropic | Google | more
OpenAI solves distance problem, Anthropic launches sandboxes, Google unveils Flash AI, Cohere releases multilingual LLM. Discover more in Today’s Global AI Native Industry Insights.
1. OpenAI model solves 80-year-old planar unit distance problem posed by Paul Erdős
An internal OpenAI model has autonomously disproved a central conjecture in the planar unit distance problem, a famous mathematical question first posed by Paul Erdős in 1946. For nearly 80 years, mathematicians believed the best solutions resembled square grids, but the AI model discovered an entirely new family of constructions that performs better. This marks the first time AI has independently solved a prominent open problem in mathematics without human scaffolding or guidance.
Read more: https://openai.com/index/model-disproves-discrete-geometry-conjecture/
Video Credit: @OpenAI on X
2. Anthropic launches self-hosted sandboxes and MCP tunnels for Claude Managed Agents
Anthropic announced two new features for Claude Managed Agents at the Code with Claude conference in London. Self-hosted sandboxes, available in public beta, allow companies to execute AI agent tools within their own infrastructure or through managed providers like Cloudflare, Daytona, Modal, and Vercel while keeping agent orchestration on Anthropic’s servers. MCP tunnels, in research preview, enable agents to securely access internal databases, private APIs, and ticketing systems through encrypted connections without exposing them to the public internet. Both features help enterprises maintain security controls and compliance requirements while using AI agents.
Read more: https://claude.com/blog/claude-managed-agents-updates
Video Credit: @claudeai on X
3. Google launches Gemini 3.5 Flash AI model with game development capabilities in Canvas
Google announced the launch of Gemini 3.5 Flash, a new AI model that enables users to build games directly from text prompts and images without complex 3D modeling. The model works within Google’s Canvas platform, allowing users to transform everyday objects into interactive digital experiences and refine gameplay through iterative prompts. Gemini 3.5 Flash delivers frontier-level intelligence at high speed, outperforming the previous Gemini 3.1 Pro model on multiple benchmarks including coding tasks. The model is now available globally through the Gemini app and represents Google’s focus on combining advanced AI reasoning with practical action-oriented capabilities.
Read more: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/
Video Credit: @GeminiApp on X
4. Cohere releases Command A+ open-source LLM with multimodal capabilities and 48-language support
Cohere announced the release of Command A+, a 218B parameter Mixture-of-Experts model with 25B active parameters available under Apache 2.0 license. The multimodal model supports text and image inputs across 48 languages and is optimized to run efficiently on minimal hardware, including a single NVIDIA Blackwell GPU. Command A+ is designed for enterprise agentic workflows with improved performance on question answering, data analysis, and memory tasks compared to previous models.
Read more: https://cohere.com/blog/command-a-plus
Video Credit: @cohere on X
That’s all for today’s Global AI Native Industry Insights. Join us at AI Native Foundation Membership Dashboard for the latest insights on AI Native, or follow our linkedin account at AI Native Foundation and our twitter account at AINativeF.