OpenAI releases an improved GPT-5.1, Anthropic targets 2028 profitability, and Cursor raises at $29.3B to build its own model.
Anthropic projects profitability by 2028 with a leaner, enterprise-focused model, while OpenAI expects $74B in 2028 losses due to heavy AI infrastructure investments and broader ambitions. Both startups face scrutiny over massive cash burns and future returns.
Cursor raised $2.3B at a $29.3B valuation to advance its AI model Composer and reduce reliance on third-party models with intensifying competition from OpenAI and Anthropic in the AI coding space.
Anthropic is investing $50B with Fluidstack to build U.S. data centers supporting Claude’s growth and frontier AI research, creating 3,200 jobs and advancing infrastructure aligned with national AI goals.
Meta AI chief Yann LeCun plans to exit and launch a startup focused on world models, amid internal turmoil and reorganization at Meta as it races to catch up in the AI arms race.
Microsoft is doubling its data center footprint with a 1M sq. ft. AI hub in Atlanta, packed with Nvidia GPUs and high-speed links to support soaring AI demand, especially for OpenAI and its own models.
Meta will begin evaluating employees on “AI-driven impact” in 2026, rewarding those who use AI to boost productivity. It’s part of a broader shift toward mandatory AI adoption across Big Tech.
ElevenLabs’ new Iconic Voice Marketplace offers licensed AI-generated voices of celebrities and historical figures, ensuring ethical use through verified rights holders and consent-based deals.
GPT‑5.1 boosts Instant with warmer, faster responses and Thinking with clearer, adaptive reasoning. Enhanced tone controls and personalization make ChatGPT smarter, more natural, and easier to tailor.
GPT-5.1 boosts speed and token efficiency with adaptive reasoning, offers a “no reasoning” mode, extended 24h prompt caching, and new tools for coding and shell command execution, while significantly improving performance across agentic and coding tasks.
Scribe v2 Realtime offers sub-150 ms multilingual speech-to-text with 93.5% accuracy, built for live agents and captioning, featuring advanced language detection and enterprise-grade compliance.
Google’s Code Wiki auto-generates live, structured documentation with interactive navigation and Gemini chat, helping developers instantly understand codebases and reduce onboarding time.
SIMA 2 combines Gemini’s reasoning with self-improvement to create a generalist AI that understands, acts, and learns across unseen 3D game worlds, advancing AI toward embodied intelligence.
Available in public beta for Sonnet 4.5 and Opus 4.1.
Besides Grok Code Remote, users will be able to run coding tasks from their local machines via CLI interface too (not available yet).
A new way to collaborate with friends, family, or coworkers and ChatGPT in the same conversation.
The evaluation tests the ideal behavior they want Claude to have in political discussions, and tests a selection of AI models for even-handedness.
A Chinese-backed group used AI autonomously to infiltrate 30+ organizations in a major cyberattack, exploiting Claude’s coding and agentic abilities. This marks the first large-scale AI-executed cyber operation with minimal human oversight.
GPT-5.1 improves on GPT-5 with faster response times, better prompt calibration, and new tools for coding and system interaction. It introduces the “none” reasoning mode, enhanced agent steerability, and refined techniques for prompt optimization in production agents.
Training neural networks to be sparse from the start produces cleaner, simpler internal circuits that are easier to understand while still performing useful tasks.
A curated weekly roundup of AI news, tools, and breakthroughs to stay ahead of the curve in a fast-moving industry.
Curated by devs, for devs. Helping 400+ developers stay ahead of the AI curve.