version 56 · published 37d ago · superseded by v59
Current state
Codex shipped 150+ updates in 2 months, including SLGO for parallel multi-agent workflows and Ultra Mode for complex project planning [6091].Codex v0.144.3 introduced mobile control via ChatGPT app, multi-agent handoffs, and DigitalOcean cloud workspace integration [6090]. OpenAI released a major Codex update enabling plugins by default, npm marketplace integration, and improved MCP tool search [6089]. OpenAI updated Codex with features including remote control via phone, DigitalOcean cloud workspaces, multi-agent task splitting, and indexed web search [6088]. Google's Gemma 4 model now uses MTP drafters to accelerate generation up to 3X locally, with the 31B model jumping from 20.8% to 89% on AIME [6087]. Kimi K3, a 2.8 trillion parameter open-weight model, was released and reportedly matches or beats GPT 5.6 Soul and Fable 5 on benchmarks, with a demonstration of 122 coding tasks from a single prompt [6085]. Spriterrific, an AI-powered sprite generation tool, is nearing release and will support agent-first integration with Cursor, Claude Code, and Codex [6084]. Every introduces All Access tier including Builder Pack with $7,000 in credits and discounts for AI tools like Codex, Cursor, and PostHog [6083]. A new plugin called Codex-Orchestration allows assigning different AI models to roles like advisor and executor within Codex [6082]. Anthropic released an official open-source skill called Frontend Design for Claude Code, enabling aesthetic-driven UI generation [6080]. Anthropic acknowledged user frustration and announced standardizing Fable 5 access at 50% usage for plans that use it most intensively, with continued investment in capacity [6078]. Anthropic launched Claude for Teachers, offering free premium Claude capabilities to verified K-12 educators in the US, with teaching skills and evidence-based curricula mapped to academic standards [6077]. Anthropic announced that from July 20, Claude Fable 5 will be included in Max and Team Premium plans at 50% limits, with Pro and Team Standard users getting usage credits and a one-time $100 credit [6076]. Claude Code received numerous fixes: removed obsolete Codex workaround [6075], preserved composer caret after request updates [6074], resynced timeline correctly after reconnecting [6073], loaded more timeline history at once [6072], fixed project picker positioning crash [6071], kept permission auto-accept state separate for each server [6070], fixed connected project picker positioning when anchor is still loading [6069], fixed spacing between model provider sections in Settings [6068], fix(core): tolerate AlreadyExists in FSUtil.ensureDir [6067], hid drawer close button on Windows [6066], stopped subagents from launching nested subagents by default with configurable subagent_depth limit [6065], home page command palette search can find and open sessions [6064], fixed custom agent selector visibility [6063], fixed startup readiness including WSL server loading [6062], improved dangerous-command detection [6061], fixed plan approvals without edits being labeled "(edited by user)" [6060], fixed hook callback timeout misreported as user rejection [6059], fixed session crash when tool result renderer returned bigint or plain text [6058], fixed paste markers leaking into external editors [6057], fixed rendered text fragment leaking into crash telemetry [6056], fixed ultracode keyword opt-in firing on non-human-originated input [6055], added startup warning for Write/NotebookEdit/Glob permission rules [6054], added live elapsed-time counter to collapsed tool summary line [6053], fixed edits that leave input as "?" being silently swallowed [6052], fixed file upload validation for DOS device suffixes and multiple hard links [6051], fixed Claude Code on Vertex and Bedrock attempting default Opus model at startup [6050], fixed auto mode overriding PreToolUse hook's ask decision for unsandboxed Bash [6049], fixed permission previews not neutralizing bidirectional-override characters [6048], fixed shell mode not executing commands containing file paths while autocomplete popup was open [6047], fixed SIGTERM during Bash tool orphaning process tree [6046], fixed continue:false hook's halt being dropped [6045], fixed plan mode auto-running file-modifying Bash commands without permission prompt [6044], added session-wide limit on WebSearch tool calls (default 200) [6043], added message.uuid, client_request_id, and tool_source to OpenTelemetry log events [6042], added ISO modified timestamp to memory file frontmatter [6041], fixed Bash permission checks for help/man commands [6040], fixed Bash permission checks for zsh variable subscripts [6039], fixed Bash permission checks for very long commands [6038], fixed Bash permission checks for file-descriptor redirect forms [6037], fixed permission-check bypass in Windows PowerShell 5.1 [6036], fixed single-segment dir/** allow rules [6035].[54]
Timeline
Codex shipped 150+ updates in 2 months, including SLGO for parallel multi-agent workflows and Ultra Mode for complex project planning [6091]. Codex v0.144.3 introduced mobile control via ChatGPT app, multi-agent handoffs, and DigitalOcean cloud workspace integration [6090]. OpenAI released a major Codex update enabling plugins by default, npm marketplace integration, and improved MCP tool search [6089]. OpenAI updated Codex with features including remote control via phone, DigitalOcean cloud workspaces, multi-agent task splitting, and indexed web search [6088]. Google's Gemma 4 model now uses MTP drafters to accelerate generation up to 3X locally, with the 31B model jumping from 20.8% to 89% on AIME [6087]. Kimi K3, a 2.8 trillion parameter open-weight model, was released and reportedly matches or beats GPT 5.6 Soul and Fable 5 on benchmarks, with a demonstration of 122 coding tasks from a single prompt [6085]. Spriterrific, an AI-powered sprite generation tool, is nearing release and will support agent-first integration with Cursor, Claude Code, and Codex [6084]. Every introduces All Access tier including Builder Pack with $7,000 in credits and discounts for AI tools like Codex, Cursor, and PostHog [6083]. A new plugin called Codex-Orchestration allows assigning different AI models to roles like advisor and executor within Codex [6082]. Anthropic released an official open-source skill called Frontend Design for Claude Code, enabling aesthetic-driven UI generation [6080]. Anthropic acknowledged user frustration and announced standardizing Fable 5 access at 50% usage for plans that use it most intensively, with continued investment in capacity [6078]. Anthropic launched Claude for Teachers, offering free premium Claude capabilities to verified K-12 educators in the US, with teaching skills and evidence-based curricula mapped to academic standards [6077]. Anthropic announced that from July 20, Claude Fable 5 will be included in Max and Team Premium plans at 50% limits, with Pro and Team Standard users getting usage credits and a one-time $100 credit [6076]. Claude Code received numerous fixes and improvements as listed in the current state.[13]
Upcoming catalysts
No upcoming catalysts are mentioned in the provided claims.
What changed in this versionvs v55
Updated to include new Codex updates (150+ updates, SLGO, Ultra Mode, v0.144.3, plugins by default, npm marketplace, MCP tool search, remote control, DigitalOcean, multi-agent splitting, indexed web search), Gemma 4 with MTP drafters, Kimi K3 release, Spriterrific nearing release, Every All Access tier, Codex-Orchestration plugin, Anthropic's Frontend Design skill, Fable 5 access standardization, Claude for Teachers, and numerous Claude Code fixes.
+Codex shipped 150+ updates in 2 months, including SLGO for parallel multi-agent workflows and Ultra Mode for complex project planning [6091].
+Codex v0.144.3 introduced mobile control via ChatGPT app, multi-agent handoffs, and DigitalOcean cloud workspace integration [6090].
+OpenAI released a major Codex update enabling plugins by default, npm marketplace integration, and improved MCP tool search [6089].
+OpenAI updated Codex with features including remote control via phone, DigitalOcean cloud workspaces, multi-agent task splitting, and indexed web search [6088].
+Google's Gemma 4 model now uses MTP drafters to accelerate generation up to 3X locally, with the 31B model jumping from 20.8% to 89% on AIME [6087].
+Kimi K3, a 2.8 trillion parameter open-weight model, was released and reportedly matches or beats GPT 5.6 Soul and Fable 5 on benchmarks, with a demonstration of 122 coding tasks from a single prompt [6085].
+Spriterrific, an AI-powered sprite generation tool, is nearing release and will support agent-first integration with Cursor, Claude Code, and Codex [6084].
+Every introduces All Access tier including Builder Pack with $7,000 in credits and discounts for AI tools like Codex, Cursor, and PostHog [6083].
−Kimi K3, a 2.8 trillion parameter open-weight model, matches Fable 5 and GPT 5.6 on benchmarks at lower cost [5890]; it is now integrated into the Codex app with a 5-minute setup [5893].
−The Claude Code team built an agent loop system of 18,000 agents, achieving 140,000 steps with 90% zero human typing [5891].
−NVIDIA released DeepStream 9.1, adding 13 agentic skills for building video analytics pipelines via natural language, compatible with Claude Code and Codex [5888].
−Anthropic released 'Frontend Design', an official skill for Claude Code that forces aesthetic direction before coding, with 161,000 stars [5886], and an official prompts library with ready-to-use prompts for development workflows [5885].
−Moonshot AI launched Kimi Code CLI, an open-source MIT-licensed CLI that clones Claude Code with additional features [5883].
−Claude for Teachers offers free premium Claude capabilities to verified K-12 educators in the US, with teaching skills and evidence-based curricula mapped to academic standards [5880].
−Codex desktop app received numerous fixes: removed obsolete Codex workaround [5879], preserved composer caret after request updates [5878], resynced timeline correctly after reconnecting [5877], loaded more timeline history at once [5876], fixed project picker positioning crash [5875], kept permission auto-accept state separate for each server [5874], fixed connected project picker positioning when anchor is still loading [5873], fixed spacing between model provider sections in Settings [5872], fix(core): tolerate AlreadyExists in FSUtil.ensureDir [5871], hid drawer close button on Windows [5870], stopped subagents from launching nested subagents by default with configurable subagent_depth limit [5869], home page command palette search can find and open sessions [5868], fixed custom agent selector visibility [5867], fixed startup readiness including WSL server loading [5866], improved dangerous-command detection [5865], fixed plan approvals without edits being labeled "(edited by user)" [5864], fixed hook callback timeout misreported as user rejection [5863], fixed session crash when tool result renderer returned bigint or plain text [5862], fixed paste markers leaking into external editors [5861], fixed rendered text fragment leaking into crash telemetry [5860], fixed ultracode keyword opt-in firing on non-human-originated input [5859], added startup warning for Write/NotebookEdit/Glob permission rules [5858], added live elapsed-time counter to collapsed tool summary line [5857], fixed edits that leave input as "?" being silently swallowed [5856], fixed file upload validation for DOS device suffixes and multiple hard links [5855], fixed Claude Code on Vertex and Bedrock attempting default Opus model at startup [5854], fixed auto mode overriding PreToolUse hook's ask decision for unsandboxed Bash [5853], fixed permission previews not neutralizing bidirectional-override characters [5852], fixed shell mode not executing commands containing file paths while autocomplete popup was open [5851], fixed SIGTERM during Bash tool orphaning process tree [5850], fixed continue:false hook's halt being dropped [5849], fixed plan mode auto-running file-modifying Bash commands without permission prompt [5848], added session-wide limit on WebSearch tool calls (default 200) [5847], added message.uuid, client_request_id, and tool_source to OpenTelemetry log events [5846], added ISO modified timestamp to memory file frontmatter [5845], fixed Bash permission checks for help/man commands [5844], fixed Bash permission checks for zsh variable subscripts [5843], fixed Bash permission checks for very long commands [5842], fixed Bash permission checks for file-descriptor redirect forms [5841], fixed permission-check bypass in Windows PowerShell 5.1 [5840], fixed single-segment dir/** allow rules [5839].
−Kimi K3, a 2.8 trillion parameter open-weight model, matches Fable 5 and GPT 5.6 on benchmarks at lower cost [5890]; it is now integrated into the Codex app with a 5-minute setup [5893].
Added Codex SSD damage reports, Kimi K3 benchmarks, Claude Final Boss open-source release, free model switching repo, Kimi Code CLI, and Fable 5 access standardization. Removed earlier Codex plugin defaults and Undo tool mentions.
38d ago
Why archives exist
every answer is versioned — nothing moves silently.