Fresh stories
Codex users report one-shot fixes and 1.7B-token days vs Claude Code
Developers posted side-by-side reports of faster one-shot fixes, 1.7B-token workdays, and fewer limit warnings with GPT-5.5 fast mode after OpenAI added Claude Code import. The comparisons matter because they turn migration talk into a concrete workflow choice.
Agent Harness Framework launches with Daytona default sandbox
The Agent Harness Framework started rolling out with Daytona as the default sandbox, and Fred Schott reported 35 pull requests on day one. The launch matters because it gives builders a packaged sandbox baseline instead of wiring execution isolation and agent environment management from scratch.
OpenClaw 2026.5.2 adds Grok 4.3 default chat and plugin fixes
OpenClaw 2026.5.2 shipped Grok 4.3 as the default xAI chat model and a broad plumbing pass for plugins, session paths, messaging bridges, and voice features. The release matters because it trims startup stalls and cleans up common integration edges in self-hosted agent setups.
Claude Code users report HERMES.md extra billing and ban appeals
Users on Hacker News and Reddit reported a reproduced HERMES.md extra-usage billing bug, plus new ban appeals and repeated blame-shifting complaints. Anthropic says affected users will get refunds and credits, so teams should keep an eye on quota routing and support escalation.
Codex users report one-shot fixes and 1.7B-token days vs Claude Code
Developers posted side-by-side reports of faster one-shot fixes, 1.7B-token workdays, and fewer limit warnings with GPT-5.5 fast mode after OpenAI added Claude Code import. The comparisons matter because they turn migration talk into a concrete workflow choice.
Practitioners report harness playbooks with Playwright CLI, create_agent, and MCP
Builders shared concrete Symphony, create_agent, and MCP setup guides after arguing that model switching is easy but harness switching is not. The playbooks matter because they make harness engineering more repeatable, so teams can copy tested tooling and integration patterns.
Developers report DeepSeek V4 Flash handles 32M-token coding runs for $0.25
Users reported moving long coding sessions from Claude to DeepSeek V4 Flash and seeing tens of millions of tokens cost only cents. Hacker News discussion also leaned toward Flash over Pro for day-to-day use, so teams should test whether the low published prices hold in their own workflows.
Agent Harness Framework launches with Daytona default sandbox
Codex adds `/hatch` pets, in-pet chat replies, and one-curl Petdex installs
OpenCode v1.14.33 fixes custom-agent loading and shifts plugins to on-demand installs
OpenClaw 2026.5.2 adds Grok 4.3 default chat and plugin fixes
Top storiesthis week
OpenAI adds one-click Claude Code migration to Codex
OpenAI added one-click import for settings, plugins, agents, and project config into Codex, and users reported cleaner workflows with visible subagents and in-chat CI status. That reduces setup friction for existing agent stacks, and OpenAI says Codex revenue doubled in under seven days.
Claude Code users report keyword-trigger billing after Opus 4.7 rollout
Days after Opus 4.7 launched, users reported commit-message triggers tied to OpenClaw or HERMES markers that could route requests into extra billing or refusals, alongside continued throttling complaints. Anthropic says affected users will get refunds, but repo-scanning heuristics may still affect cost and reliability in multi-harness workflows.
Moondream releases Photon 1.2.0 with Apple Silicon, native Windows CUDA, and 23 ms B200 latency
Moondream shipped Photon 1.2.0, expanding its inference engine to Apple Silicon, Windows CUDA, Blackwell, and Jetson Thor, then outlined how custom Metal kernels and fused ops made local vision practical without MLX. That broadens deployment options for edge and on-device vision workloads while keeping server-class latency on B200 systems.
ARC Prize reports GPT-5.5 at 0.43% and Opus 4.7 at 0.18% on ARC-AGI-3
ARC Prize published frontier-model results on ARC-AGI-3 and said GPT-5.5 and Opus 4.7 both stayed below 1%, with failures in world modeling, abstraction, and reward reinforcement. That shows strong coding and benchmark models still break on novel interactive reasoning tasks, and follow-up comparisons even had Opus 4.6 slightly ahead of 4.7.
OpenClaw adds ChatGPT subscription sign-in for hosted coding-agent sessions
OpenClaw now lets users sign in with a ChatGPT account and use an existing subscription inside the harness. That gives teams a sanctioned access path for third-party agent workflows just as Claude Code users report tighter billing and refusal heuristics elsewhere.