Ultraplan hands your planning task to a dedicated cloud session running Opus 4.6 for up to 30 minutes — while your terminal stays free. Here’s what it actually is, how the three modes differ, and when to reach for it.
SWE-bench Verified — the benchmark that put every frontier model above 80% — is contaminated. OpenAI stopped reporting it in February. Here’s what actually happened, what SWE-bench Pro replaces it with, and why 46% is a more honest number than 81%.
Cursor 3 shipped a genuinely redesigned interface built around parallel agents. The Agents Window, Design Mode, /worktree, and /best-of-n are real features with real uses. But ‘agent-first’ describes the UI layer, not the architecture — and the distinction matters more than Cursor’s marketing suggests.
Anthropic launched Claude Managed Agents on April 8 — a managed API that handles the agent loop, sandboxing, checkpointing, and tool orchestration you’d otherwise build yourself. Here’s what it actually offers, how the pricing model works, and why it matters for teams shipping production agents.
Meta Superintelligence Labs shipped Muse Spark — and made it closed-source. The company that framed open AI as a moral imperative just locked the door. Here’s what that means for developers who built their stack on Llama.
GitHub Copilot’s April 8 VS Code update ships Autopilot Mode, nested subagents, and MCP sandboxing. These are real improvements. They’re also a demonstration of why bolting autonomy onto an IDE produces something fundamentally different from a real agent.
Z.AI released GLM-5.1 today — a 754B open-weight model under MIT license that scored 58.4% on SWE-bench Pro, beating GPT-5.4, Claude Opus 4.6, and Gemini 3.1 Pro. Its headline demo: an 8-hour autonomous session that built a complete Linux desktop environment across 655 iterations. The closed-model monopoly on frontier coding capability just got its first serious challenge.
Anthropic officially unveiled Claude Mythos Preview on April 7, confirming what the March leak hinted at: a model that autonomously found thousands of zero-days across every major OS and browser. Their response — Project Glasswing — grants restricted access to a select group of tech giants to use Mythos as a defensive weapon. This is the most consequential ’too dangerous to release’ moment in AI history.
A patched vulnerability in Claude Code (CVE-2026-21852) reveals an entirely new attack surface: poisoned project config files that silently bypass your deny rules and exfiltrate credentials. Here’s what happened, how the exploit works, and what it means for agentic security.
Spec-Driven Development has crossed from niche methodology to recognized category — with 30+ competing frameworks, a conference track at Agentic Conf Hamburg, AWS Kiro as the first commercial SDD IDE, and enterprise backing from McKinsey and Anthropic’s own trend reports. Here’s what’s happening and what it means.