Wednesday, July 22, 2026

Models, benchmarks, the lab race. · 5 stories

1. 1Cisco Foundation AI Releases Antares: 350M and 1B Open-Weight Models That Localize Known Vulnerabilities Inside Real Codebases78
2. 2Cisco bets its small open cybersecurity models can outperform GPT-5.5 at vulnerability detection for a fraction of the cost
3. 3Unsloth vs Axolotl vs TRL vs LLaMA-Factory: A Fine-Tuning Framework Comparison on Speed, VRAM, and Multi-GPU

The AI products people actually use. · 20 stories

1. 1OpenAI introduces Presence to help enterprises build AI agents78
2. 2Cursor Releases Cursor Router: A Request-Level Classifier Delivering Frontier Coding Quality at 30–50% Lower Cost
3. 3Launching Health in ChatGPT

Autonomous systems in the wild. · 0 stories

1. Nothing cleared the bar today.

Funding, M&A, valuations, earnings. · 18 stories

1. 1Anthropic will deploy 2 gigawatts of AMD GPUs for Claude in a deal worth up to $5 billion92
2. 2AMD commits up to $5 billion to Anthropic
3. 3AMD to invest up to $5 billion in Anthropic as part of computing power deal

Silicon, racks, the compute buildout. · 14 stories

1. 1Anthropic to buy up to 2 gigawatts of GPU capacity from AMD92
2. 2OpenAI’s AI spending spree has ballooned to $750B
3. 3Google burns through $6bn in cash as AI spending climbs again

Jobs, industries, people & policy. · 28 stories

1. 1Every frontier AI model tested by Britain's safety institute tried to cheat on cybersecurity evaluations78
2. 2Treasury threatens sanctions after White House claims Moonshot distilled Anthropic’s Fable
3. 3America Says Its AI Has No Kill Switch, But The World Is Hedging