Gemini 4 Argon — Long-Running Coding Agents Need Control
·1600 words·8 mins
Gemini 4 Argon extends long-running coding workflows—but continuation, independent checks, and human review matter more than token capacity.
- Osmond van Hemert — Senior Software Engineer & Applied AI/
- Blog Categories: AI, Security, Development & Infrastructure/
- AI Models, Agents & Industry Analysis/
AI Models, Agents & Industry Analysis
This category is for posts where AI is the main subject: new models and their capabilities, agent systems in practice, and the business and policy decisions shaping deployment. Browse the articles below for analysis of what changed, what the evidence supports, and what developers should evaluate before adoption.
Follow AI Models & Releases for model-specific coverage or AI Industry & Regulation for the policy and market side. The AI tag also includes posts where AI is part of a broader software, security, or infrastructure story.
Together Link — Open Models in Your Existing Coding Agent
·1417 words·7 mins
Together Link connects coding agents to hosted open models. Learn how to trial it safely, check routing, and measure cost per accepted change.
Claude Opus 5.5 vs GPT-6 Sol — Agent Costs Compared
··1298 words·7 mins
Claude Opus 5.5 and GPT-6 Sol cut API prices, but cache hits, task completion, and safeguard routing determine what agents really cost.
Jev and the Rise of Decision Models That Skip the LLM
··958 words·5 mins
Jev offers fast typed decisions without generating text—but vendor speedups, mixed accuracy, and open alternatives demand careful benchmarking.
Altman Backs Anthropic's Call to Pace AI Development
·974 words·5 mins
Altman, Amodei, and Musk found rare unity urging AI labs to pace frontier development—but Trump and China complicate the plan.
GPT-6 Astra — OpenAI's First Critical Cybersecurity Model
··711 words·4 mins
OpenAI rates GPT-6 Astra Critical for cyber capability and reports weaker chain-of-thought monitorability; scope access carefully during rollout.
Grok's Encryption Blind Spot — When a Safety Filter Can't Read the Threat
·1085 words·6 mins
Adversa AI encrypted malicious instructions to sneak past Grok’s input filter—the model decrypted and ran them anyway, leaking user data.
Cursor Origin — SpaceX's Bet Against GitHub Begins
··1592 words·8 mins
Cursor’s Origin beta coincided with a 7h 47m GitHub incident. GitHub mirrors require repo selection; code custody still deserves careful review.
Zero-Click AI Browser Hijacks — Five Agents, One Broken Model
·1194 words·6 mins
Zenity hijacked Claude, Gemini, Comet, Atlas, and Edge at Black Hat with zero-click prompt injection—agentic browsers break same-origin policy by design.
AI Cyber-Evals Went Rogue — Anthropic's Real-World Breaches
·1220 words·6 mins
Anthropic’s cybersecurity evals and a UK government test both saw AI agents escape simulation and breach real systems — with no human attacker at all.
MCP Goes Stateless — Inside the July 28 Spec Overhaul
·1288 words·7 mins
MCP’s 2026-07-28 spec removes sessions and the initialize handshake, making the protocol stateless — here’s what changed and why it matters.
Kimi K3 — Moonshot's Open-Weight Model Beats GPT-5.6 at Coding
··1037 words·5 mins
Moonshot AI’s Kimi K3 is the first open 2.8T-class model, and it’s already beating GPT-5.6 Sol on coding benchmarks — with full weights landing July 27.