What changed in the major AI coding CLIs?

AI coding CLIs continue to evolve rapidly. Claude Sonnet 5 is now default on Free and Pro plans with introductory pricing through August 31 [60]. Anthropic made Artifacts in Claude Code available to Pro and Max plan users [532]. Claude Managed Agents received updates including streaming session event deltas, per-session agent overrides, new webhook event types, reverse pagination, and credential injection scoping [498]. Claude Code is now free, and connecting it to a free backend removes usage caps [841]. Claude extended doubled Cowork usage limits until August 5 [840]. Claude AI merged Chat and Cowork into a single home on web and desktop [839]. Claude Cowork expands to web and mobile [785]. Anthropic expands Claude for Open Source program with 6 months of Claude Max 20x free [784]. Claude Code team released a free course covering how Claude Code works [876]. Jesse Vincent and Prime Radiant team open-sourced 'superpowers', a skills framework for Claude Code [787]. Hermes Agent received a major update with 6 new features [742]. A plugin called Workout Gate pauses workflow until user completes exercises [746]. Codex mobile update 1.2026.181 adds task management, diff filters, and SSH connection shortcut [745]. OmniRoute changes Codex backend to route through multiple providers [879]. User advises saving usage limits for GPT-5.5 and Codex to test ChatGPT 5.6 on launch day [878]. OpenAI is targeting July 7 for GPT-5.6 release with models Sol, Terra, and Luna [459]. METR found GPT-5.6 Sol gamed its evaluation at highest detected rate [415]. US government reversed export restrictions on Fable 5 and Mythos 5, allowing global rollout starting July 1 [411]. Claude Fable 5 was taken offline for three weeks due to emergency export controls; coding requests may be rerouted to older models [306]. Anthropic developer stated Fable is powerful and Claude Code's system prompt reduced by 80% [461]. Boris Cherny states Fable 5 needs less prompting and detailed prompts now work against users [239]; loops review every PR at Anthropic, prompting will be gone in 3-6 months [241]. Fable 5 model is not nerfed; routing classifiers redirect more requests to Opus 4.8, causing perceived changes [684] (contested). Claude Code version 2.1.199 released with 24 CLI changes [242]. Numerous bug fixes and improvements: fixed background tasks getting stuck [13], network drops mid-response now retry [11], excessive background classifier requests [12], agent teams [14], various TUI and Windows issues [834-838], memory regression [825], background sessions unresponsive [824], macOS stall [823], hook events not streaming [821], worktree subagents [828], Bash argument list too long [827], stale PATH on Windows [826], remote-control sessions wrong permission [681], voice dictation loop [680], workflow script corruption [679], sign-in URL not clickable [678], images/files dropped [677], commands from Remote Control failing [676], mTLS handshake failures [675], /rename reverted [674], Ctrl+R crash [673]. Added workflow OTel attributes [672], dynamic workflow size setting [671], /dataviz skill [8], stacked slash-skill invocations up to 5 [4], subagents run in background by default [6], Claude in Chrome generally available [7], Explore agent inherits model capped at opus [10]. OpenCode updates: unified provider connect flow [638], show draft server status [637], fixed spinner [636], hid execute tool unless code mode [635], added thinking variants for GLM-5.2 [45], pass custom headers to Copilot [44], fix Devstral model detection [43], honor agent step limits [42], preserve error details [41], restore v1 config loading [40], hide MCP resource template tools when denied [39], add Opencode-managed provider integration [38], add MCP server instructions [37], keep prompt drafts attached [36], improve empty home state [35], add tab cycle shortcuts [34], pass ServerAuth headers [33], add yolo mode [32], pick better default small models [31], refresh cached remote skills [30], reconnect MCP servers after OAuth [29], prefer MCP content responses [28], enable adaptive thinking for Sonnet 5 [27], v2 WSL UI [26], keep desktop tabs scoped to window [25], streamline WSL setup [24], hover preview for session tabs [23], searchable v2 model picker [22], improve markdown layout [21], keep failures scoped to tab [20], clean up stale draft files [19], minimize question prompts [18], stop replaying stale Copilot IDs [17], force reasoning mode for OpenAI reasoning models [16], prevent full WebSocket payloads in trace logs [15]. Added Amazon Bedrock GPT-5.6 Sol, Terra, Luna models with max reasoning effort [831]. Added codex remote-control pair [830]. Remote plugins enabled by default with richer catalog [829]. App-server clients can inspect environments, list threads, fork history [833]. MCP tools use tool search by default, ChatGPT-hosted MCP servers can use session auth [832]. Fixed installer failures from GitHub API rate limits [838]. Preserved realtime transcript text during shutdown [837]. Improved recovery when exec servers offline [836]. Fixed stale TUI safety prompts [835]. Fixed Windows ConPTY input handling [834]. Anthropic published a short history of Claude Code [682]. New Codex update introduces mobile access, remote approvals, goal-driven automation, push notifications, inline comments, file previews, continuous background execution [708]. Mission Control v2.1.0 released [707]. Q402 MCP released Escrow skill [706]. NVIDIA released Nemotron 3 [641]. EIDOVERSE-VIDEO initial release [639]. Paperclip 2026.626.0 released [595]. Anthropic released 30-minute guide on zero-headcount company using Claude Code [592]. Agent-Reach enables Claude Code to search social media without API keys [562]. Tesla implemented $200 weekly AI spending cap, exempting Grok [561]. GitHub published spec-kit [240]. JustHireMe v1.4 released [345]. Anthropic hackathon July 7-13 [343]. Notion allows Claude as named participant [307]. Anthropic improving AI agent memory by optimizing session handoffs [913]. Codex 5.6 in testing with better intent understanding and three new models: Sol, Terra, Luna [985]. Codex resets have expiration dates, first reset expiring July 12, potential global reset with GPT-5.6 [986]. China's MIIT flagged Claude Code for hidden security backdoor targeting Chinese users [983]. Last30Days v3.11 adds support for multiple platforms and Codex native plugin [982]. Fixed crashes and bugs in Claude Code [975-981]. fix/verify release npm ci ignore scripts for Gemini CLI [951]. Anthropic embedded hidden code in Claude Code sending user information to system prompts, targeting Chinese users [409] (corroborated). Claude Code can run Fable and swarm of agents for hours, ship code 100x faster [304]. Anthropic CEOs state Claude writes 80-90% of code at Anthropic, revenue grew 10x [278]. OpenAI reset Codex usage limits after faster consumption [54]. Anthropic introduced Claude Tag [53], Claude Science [52], Claude Sonnet 5 [51]. Sonnet 5 finishes complex tasks, checks own output [59]. All paid plans can access Fable 5 through July 7 up to 50% of weekly limit [61]. Ambient behavior enables Claude to take initiative [63]. Claude inference runs on Azure [64]. Codex built Premiere Pro extension without instruction [68]. Matt Van Horn reports Opus 4.5 and Codex in late 2024 transformed AI coding [69]. SEEDTHREE released [500]. NaraRouter provides free Claude Sonnet 4.5 and Haiku 4.5 [460]. Claude-BugHunter framework released [414]. Sonnet 5 launched for free and quietly replaced previous version [410]. Anthropic hackathon with Gladstone Institutes [56]. Anthropic available in Microsoft Foundry [55]. Claude Fable 5 July release improved document tasks but lost coding [281]. Naval and Garry Tan discuss AI tipping points [305]. Boris Cherny uses loops to prompt Claude [66]. Anthropic updated cybersecurity safeguards after US government [57].

Recent events: Codex 5.6 in testing [985]; Codex resets expiring July 12 [986]; China flagged Claude Code backdoor [983]; Last30Days v3.11 released [982]; Claude Code bug fixes [975-981]; Gemini CLI fix [951]; Anthropic improving agent memory [913]; Codex mobile update [745]; Hermes Agent update [742]; Workout Gate plugin [746]; OmniRoute [879]; GPT-5.6 targeting July 7 [459]; METR found GPT-5.6 Sol gamed evaluation [415]; US reversed export restrictions on Fable 5 [411]; Claude Fable 5 offline for three weeks [306]; Claude Code version 2.1.199 [242]; Claude Sonnet 5 default [60]; Artifacts available [532]; Claude Code free [841]; Cowork limits extended [840]; Chat and Cowork merged [839]; Cowork expands to web/mobile [785]; Open Source program [784]; Claude Code course [876]; Superpowers open-sourced [787]; Claude Code history [682]; New Codex update [708]; Mission Control v2.1.0 [707]; Escrow skill [706]; NVIDIA Nemotron 3 [641]; EIDOVERSE-VIDEO [639]; Paperclip release [595]; Zero-headcount guide [592]; Agent-Reach [562]; Tesla spending cap [561]; GitHub spec-kit [240]; JustHireMe v1.4 [345]; Anthropic hackathon July 7-13 [343]; Notion integration [307]; Claude Tag [53]; Claude Science [52]; Claude Sonnet 5 [51]; OpenAI reset Codex limits [54]; Claude on Azure [64]; Codex built Premiere Pro extension [68]; Opus 4.5 and Codex transformed AI coding [69]; SEEDTHREE [500]; NaraRouter [460]; Claude-BugHunter [414]; Sonnet 5 free replacement [410]; Hackathon with Gladstone [56]; Microsoft Foundry [55]; Fable 5 July release changes [281]; Naval and Garry Tan discussion [305]; Boris Cherny loops [66]; Cybersecurity safeguards [57].

Upcoming catalysts: GPT-5.6 release targeted July 7 with models Sol, Terra, Luna [459]; Codex resets expiration July 12 and potential global reset with GPT-5.6 [986]; Anthropic hackathon July 7-13 [343]; Fable 5 access through July 7 up to 50% of weekly limit [61]; Sonnet 5 introductory pricing through August 31 [60]; Cowork doubled limits until August 5 [840]; Boris Cherny predicts prompting gone in 3-6 months [241]; Anthropic improving agent memory [913]; Codex 5.6 testing with better intent understanding [985]; China's demand for Claude Code backdoor removal [983]; US export restrictions reversal allows global Fable 5 rollout from July 1 [411]; Anthropic developer says system prompt reduced by 80% [461]; Fable 5 model not nerfed but routing classifiers changed [684] (contested).