AlphaSignal flagged an OpenAI Codex run that finished a nine-hour coding task after exhausting its usage limit, using banked resets and active-turn continuation. Here is how to plan autonomous agent sessions without losing momentum.
Greptile's TREX layer runs PR branches in sandboxes and attaches logs, screenshots, and traces to review comments. Here is what that means for teams shipping with Cursor, Claude Code, and other agentic coding tools.
A stealth coding model with a 1M-token window landed on OpenRouter August 20 with zero lab name attached. Community forensics point at Zhipu GLM infrastructure, and that raises real routing questions for production code.
Google shipped Gemini 3.7 Flash three weeks after 3.6 with sharper coding, document reasoning, and web dev scores. Intro pricing is $0.75 per million input tokens. Here is how it fits an agent routing stack.
Gemini 3.7 Flash ships three weeks after 3.6 with stronger coding and agent benchmarks, intro pricing at $0.75 per million input tokens, and a path into Antigravity and Gemini Spark.
Anthropic's unreleased Claude did not solve the Riemann hypothesis. It did coordinate ~60 Claude Code subagents, run 2,400 shell commands, and ship a Lean 4 proof that raises the proven zero bound to 67.2%.
Spotify open-sourced Xirp, a vendor-neutral workspace for Claude Code, Codex, and Gemini CLI with shared context across 36,000+ sessions. Here is what that means for teams drowning in parallel agents.
Anthropic made auto mode the default on Pro, Max, and Team plans after a 1,053-tester study. The classifier blocked 89% of dangerous shell commands while manual approval fatigue dropped human catches to 5%.
Cross-session messaging in Claude Code v2.1.224 lets independent terminals share plain-text notes locally. Here is how ListAgents and SendMessage work, what stays off the wire, and when I still use agent teams instead.
Meta shipped Muse Code, a terminal coding agent with parallel git worktrees, a replay-exact event log, and Muse Spark 1.2 co-trained on the harness. It lands second on Terminal-Bench at roughly a quarter of frontier token prices. Here is what is real and what is marketing.
Alibaba's Qwen3.8-Max packs 95B active parameters into a 2.4T MoE stack, ranks ahead of Claude Fable 5 on WebDev Arena, and prices at $2/$6 per million tokens. Weights hit Hugging Face next week.
Cursor Router routes each coding request to the right model with Intelligence, Balance, and Cost modes. Early enterprise traffic saved 30-50% versus Opus 4.8 defaults with no quality drop.
Anthropic published two Claude Managed Agents patterns that keep Fable 5 as the brain while Sonnet 5 does the token-heavy work. On benchmarks, that lands near frontier quality at 46% to 63% of solo Fable cost.
Kimi K2.7 Code is the first open-weight model in Copilot's picker. I break down cost, enterprise policy defaults, and when I would route agent loops to it instead of frontier models.
pxpipe is a local proxy that renders dense system prompts and old history as PNGs before they hit Claude Code. Real workloads report 59 to 70 percent lower bills, with a lossy catch you need to understand.
Brian Armstrong's team runs 1,200 AI agents, defaults to GLM-5.2 and Kimi K2.7, and automated model selection. Five levers any engineering org can copy without a crypto-scale budget.
Z.ai's MIT-licensed GLM-5.2 ships 1M-token context, beats GPT-5.5 on several coding benches, and costs a fraction of Opus. Here's what production tests show, where it still breaks, and how I'd route it.
A privacy-preserving analysis of roughly 400,000 Claude Code sessions shows lawyers, managers, and finance folks succeed at nearly the same rate as software engineers. Verified success doubles from novice to expert, and average task value rose 27% in six months.
Z.ai's GLM-5.2 brings a 1M-token context window, IndexShare sparse attention, and MIT-licensed weights. One day after the Fable 5 export control shock, open long-horizon coding got a lot more interesting.
Moonshot's open-weight Kimi K2.7 Code keeps the 1T MoE backbone but cuts thinking tokens ~30% versus K2.6 while jumping +21.8% on Kimi Code Bench v2. Here is when I would route agents to it.
Cohere's 30B MoE North Mini Code runs on a single H100 with Apache 2.0 weights. Strong SWE-Bench numbers, 2.8x throughput, and a verbosity tax you should measure before you swap APIs.
Anthropic's Fable 5 tops CursorBench at 72.9% but costs about twice Opus 5. Here is when I route hard agentic work to Fable and when I keep Composer on the loop.
Cognition's FrontierCode benchmark grades mergeability, not just test passes. On the hardest Diamond tier, Claude Opus 4.8 leads at 13.4%. Here is how I read that number for production agent routing.
Moonshot's Kimi Work desktop agent reads local files, drives your real browser, and spins up to 300 sub-agents per task. Here is how Agent Swarm compares to cloud-only coding agents I deploy for clients.
OpenAI shipped Codex Computer Use on Windows with ChatGPT mobile remote control. Here is what foreground takeover means for testing, security, and how I would wire it into a real agent workflow.
OpenAI shipped Codex computer use on Windows with mobile steering. Here is what foreground takeover means for QA, privacy, and how I would wire it into a real dev loop.
Cursor 3.6 shipped Auto-review with a three-stage filter and a classifier subagent. Here is how it cuts terminal prompts by roughly 84% and what I configure on client machines.
Cursor 3.6 shipped auto-review on May 29, 2026 with a classifier subagent, sandbox layer, and roughly 84% fewer approval prompts. Here is how to configure it without treating convenience as a security boundary.
xAI opened grok-build-0.1 on the public API at $1 input and $2 output per million tokens. Here is how it compares for agent loops, cached context, and when I would route to it.
grok-build-0.1 hit the xAI API on May 29, 2026 at $1 per million input tokens with a 256K context window and native tool use. Here is how it fits next to Composer and frontier tiers in a real harness.
Opus 4.8 keeps Opus 4.7 pricing while adding dynamic workflows (up to 1,000 subagents), cheaper fast mode, and effort control. The real win is fewer silent failures when you walk away from a long agent run.
Datacurve's DeepSWE benchmark uses 113 original long-horizon tasks and hand-written verifiers so GPT-5.5 leads by 16 points where older SWE tests looked tied. Here is why that matters for model pickers.
multica-ai's andrej-karpathy-skills repo distills four behavioral rules from Karpathy's LLM coding critiques into a single file for Claude Code and Cursor. Here is why minimal beats another plugin marketplace.
Bearly AI's OpenADE adds a reviewable plan step before Claude Code or Codex touches your repo, with git snapshots on every run. Here is when that loop beats firing agents straight at code.
Alibaba's Qwen3.7 Max brings a 1 million token context window, strong SWE-Bench scores, and Anthropic-compatible APIs to agent workloads. Here is what that means if you ship coding agents for a living.
Composer 2.5 scores 62 on the Coding Agent Index for $0.07 per task while Opus 4.7 costs $4.10. For agent loops, routing beats defaulting to frontier models.