> ## Content Index > Fetch the complete content index at: https://www.thenewway.ai/llms.txt > Use this file to discover other available public pages before exploring further. # Hit Codex limits? OpenAI vows relief: an upgrade or a reset every day for 28 days - URL: https://www.thenewway.ai/hit-codex-limits-openai-vows-relief-an-upgrade-or-a-reset-every-day-for-28-days/ - Published: 2026-10-05T15:28:55.000Z - Updated: 2026-10-05T15:28:55.000Z - Description: OpenAI's Codex team says each day brings either an upgrade or a full usage reset. Nothing has shipped under it yet. Plus default spending caps, Copilot's retired models and Claude Code's mod patches. - Author: The New Way Tibo Sottiaux of OpenAI's Codex team says that for the next 28 days they will ship either a clear improvement or a full usage reset every day; nothing had shipped under the pledge by Monday morning. Simon Willison wants AI agents to come with a hard spending cap switched on by default. GitHub retired four Copilot models on Friday, Claude Opus 4.7 among them. And Claude Code needed two point releases in two days to steady mods, the plugins it added Thursday. 📌 **In this issue** • [OpenAI vows a Codex upgrade or a usage reset every day for 28 days](https://www.thenewway.ai/hit-codex-limits-openai-vows-relief-an-upgrade-or-a-reset-every-day-for-28-days/#openais-codex-team-vows-an-upgrade-or-a-usage-reset-every-day-for-28-days) • [Willison wants spending caps on by default](https://www.thenewway.ai/hit-codex-limits-openai-vows-relief-an-upgrade-or-a-reset-every-day-for-28-days/#simon-willison-wants-ai-spending-caps-switched-on-by-default) • [GitHub retires four Copilot models](https://www.thenewway.ai/hit-codex-limits-openai-vows-relief-an-upgrade-or-a-reset-every-day-for-28-days/#github-retires-four-copilot-models-claude-opus-47-among-them) • [Claude Code patches its new mods twice in two days](https://www.thenewway.ai/hit-codex-limits-openai-vows-relief-an-upgrade-or-a-reset-every-day-for-28-days/#claude-code-patches-its-new-mods-twice-in-two-days) --- ## What changed this week ### OpenAI's Codex team vows an upgrade or a usage reset every day for 28 days Tibo Sottiaux of OpenAI's Codex team [said on X Sunday](https://x.com/thsottiaux/status/2106845241357824205?ref=thenewway.ai) that for the next 28 days, "each day we’ll either ship one thing that is a clear improvement and relevant for most codex/work users or ship a full reset," meaning a reset of usage limits. The post passed 5 million views by Monday. [Sixteen hours earlier he said](https://x.com/thsottiaux/status/2106610099720720811?ref=thenewway.ai) the team was "locking in" on "simplifications, more efficiency for more usage, groundbreaking features or new models." To a user who called it "28 days of fixes," he replied: "Not fixes. Significant improvements." As of Monday morning, nothing has shipped under it. Last week he promised a usage reset for every paid ChatGPT account on Friday at "10am PST," and it [landed about four hours late](https://x.com/thsottiaux/status/2106131810921136451?ref=thenewway.ai). > Over the next 28 days, each day we’ll either ship one thing that is a clear improvement and relevant for most codex/work users or ship a full reset. Let the improvements begin. [https://t.co/0oFXZ7LY5T](https://t.co/0oFXZ7LY5T?ref=thenewway.ai) > > — Tibo (@thsottiaux) [October 4, 2026](https://x.com/thsottiaux/status/2106845241357824205?ref%5Fsrc=twsrc%5Etfw&ref=thenewway.ai) --- ### Simon Willison wants AI spending caps switched on by default Simon Willison, a co-creator of the Django web framework, argues that pay-by-use AI services should ship with a hard spending cap on. Coding agents, he writes, "greatly reduce the friction of spinning up code that can do useful things," which makes an overnight bill of "several hundred (or several thousand) more dollars" easy. AWS began rolling out spending limits in September to a limited set of customers, and Google Cloud added Spend Caps in July. Turning a cap off, he says, should mean ticking "I will be responsible for subsequent charges." [We’re going to need default hard budget caps on pretty much everythingHere’s a product feature which the world is going to need a whole lot more of over the coming months and years: default hard budget caps. I’m talking about the …![](https://storage.ghost.io/c/9c/1f/9c1ff71c-2859-44c2-a962-1337982a6484/content/images/icon/favicon-07f06b63-b94e-440c-9910-5949238f74ba.ico)Simon Willison’s WeblogSimon Willison](https://simonwillison.net/2026/Oct/3/default-hard-budget-caps/?ref=thenewway.ai) --- ### GitHub retires four Copilot models, Claude Opus 4.7 among them GitHub retired four Copilot models Friday: Gemini 3.5 Flash, Gemini 3.6 Flash, Kimi K2.7 Code and Claude Opus 4.7\. Their replacements are Gemini 3.8 Flash, Kimi K3 and Claude Opus 5.5\. GitHub removes the old models itself but asks you to "update your workflows and integrations to use supported models." [In a second post the same day](https://github.blog/changelog/2026-10-02-copilot-code-review-api-support-and-new-default-effort-level/?ref=thenewway.ai), it said Copilot code review now defaults to its Balanced effort level and accepts review requests through GitHub's REST and GraphQL APIs. [Selected models in GitHub Copilot deprecated - GitHub ChangelogAs of today, October 2, 2026, we have deprecated the following models across all GitHub Copilot experiences (including Copilot Chat, inline edits, ask and agent modes, and code completions). Model…![](https://storage.ghost.io/c/9c/1f/9c1ff71c-2859-44c2-a962-1337982a6484/content/images/icon/cropped-github-favicon-512-b5d24487-f0a1-47d8-8bdf-237eca809343.png)The GitHub BlogAllison](https://github.blog/changelog/2026-10-02-selected-models-in-github-copilot-deprecated/?ref=thenewway.ai) --- ### Claude Code patches its new mods twice in two days Claude Code added mods on Thursday: plugins that can redraw its interface or step into a tool call before it runs. Two point releases followed, 2.1.288 on Friday and 2.1.289 on Saturday, and most of their fixes are about keeping mods and plugins stable, such as "installed mods not loading in the first session after an upgrade." 2.1.289 also closes a sandbox gap where a Bash deny or ask rule was skipped if a bare variable assignment came before the command. [Claude Code changelog - Claude Code DocsRelease notes for Claude Code, including new features, improvements, and bug fixes by version.![](https://storage.ghost.io/c/9c/1f/9c1ff71c-2859-44c2-a962-1337982a6484/content/images/icon/android-chrome-192x192-5d70d019-1776-4867-82c2-7aa442ccf00a.png)Claude Code Docs![](https://storage.ghost.io/c/9c/1f/9c1ff71c-2859-44c2-a962-1337982a6484/content/images/thumbnail/image-d002d51e-8aa7-4b74-aa28-38e45cc0581d.png)](https://code.claude.com/docs/en/changelog?ref=thenewway.ai) --- ## Also worth your time - [AgentCraft runs a team of Claude agents inside Minecraft that plan and build in real git worktrees; a developer's demo, 993K views on X (X)](https://x.com/BlendiByl/status/2106226258984264042?ref=thenewway.ai) - [Show HN: Pi pod runs Pi, Earendil's open-source terminal coding agent, in sandboxes on your own server (HN)](https://pipod.dev/?ref=thenewway.ai) - [Show HN: Offrun gives you one workspace to manage every coding agent you're running at once (HN)](https://offrun.dev/?ref=thenewway.ai) - [One developer's month coding with Z.ai's GLM 5.3 Flash model, including where it broke (wagtail.org)](https://wagtail.org/blog/one-month-on-glm-53-flash/?ref=thenewway.ai) - [A practitioner's case for Markdown-based agent memory over RAG and vector databases — the author's own open-source tool is one example, not the only one (liao.gg)](https://liao.gg/blog/agents-dont-need-memory?ref=thenewway.ai) --- Know someone who'd want this in their inbox? Forward it — that's how this grows. And if we got something wrong, or you think we buried the real story today, hit reply. A person reads every one. --- *The New Way is written with AI. It gathers the day's stories, checks them against their sources and drafts every summary. A person decides what runs and reviews every issue before we hit send.*