Kimi K2.6: Open-Weight #1 on SWE-Bench Pro
Kimi K2.6 tops SWE-Bench Pro at 58.6%, beats GPT-5.4 and Claude. Open-weight, MIT-licensed. Here's what 300-agent swarms and 12-hour coding sessions actually mean.
Sekai Raises $26M: 15 Million AI Mini Apps and Vibe-Coding Goes Consumer
Sekai raised $26M as users built 15 million AI-generated mini apps at 200K per day. ...
AI-Generated Code Introduces Security Vulnerabilities 10x Faster
New research shows AI coding tools introduce security vulnerabilities at 10x the rate of human ...
Your AI Agent Is the Threat: 344 Enterprise Disasters
New research finds AI agents caused 344 enterprise disasters without any external attacker. Here is ...
Supabase Raises $500M: AI Agents Now Deploy Most Databases
Supabase raised $500M at $10.5B. The real story: AI agents now deploy the majority of ...
UK Police Halt AI in Court Statements — Broken Fix
UK police were ordered to stop using AI for court statements after hallucinations surfaced in evidence. The fix? Check everything it produces. Here ...
Cloudflare AI Gateway Spend Limits: Control AI Costs Now
Cloudflare AI Gateway adds dollar-based spend limits in open beta. Set budgets per user, model, ...
Koog 1.0: JetBrains Releases Stable AI Agent Framework for Java and Kotlin
JetBrains shipped Koog 1.0, the first stable AI agent framework for Java and Kotlin, with ...
Anthropic Ends Agent SDK Subsidy June 15: Act Now
Anthropic separates Agent SDK billing from subscriptions on June 15. Credits run $20-$200/month. Here is ...
Claude Opus 4.8: Why the 69.2% Number Matters More Than 88.6%
Claude Opus 4.8 is out. SWE-bench Pro: 69.2% vs GPT-5.5’s 58.6%. Reliability fixes, dynamic workflows, ...