--- title: OpenAI teams presenting at this week's conference, with... description: OpenAI teams including those behind ChatGPT, Codex, and OpenClaw are attending a conference this week. Jason Liu will speak after Romain Huet's keynote... canonical: https://daily.dev/posts/openai-teams-presenting-at-this-week-s-conference-with-codex-updates-and-monorepo-talk-on-the-agend-djuu5hwoq twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: OpenAI teams presenting at this week's conference, with Codex updates and monorepo talk on the agenda | daily.dev og:description: OpenAI teams including those behind ChatGPT, Codex, and OpenClaw are attending a conference this week. Jason Liu will speak after Romain Huet's keynote... og:url: https://daily.dev/posts/openai-teams-presenting-at-this-week-s-conference-with-codex-updates-and-monorepo-talk-on-the-agend-djuu5hwoq og:image: https://api.daily.dev/og/posts/djUu5HWoQ.png og:image:alt: OpenAI teams presenting at this week's conference, with Codex updates and monorepo talk on the agenda og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # OpenAI teams presenting at this week's conference, with Codex updates and monorepo talk on the agenda **[Collections](https://daily.dev/sources/collections)** · 6 min read · 1 upvotes · 0 comments ## Summary OpenAI teams including those behind ChatGPT, Codex, and OpenClaw are attending a conference this week. Jason Liu will speak after Romain Huet's keynote covering what's new in Codex and its roadmap. Liu is also running a casual session on his personal monorepo setup and discussing the direction of the app space. ## Content ## What launched OpenAI released GPT-5.6 alongside ChatGPT Work, a unified agentic platform that folds together ChatGPT, Codex, and a built-in browser into one app. The model comes in three tiers: Sol (flagship, complex reasoning), Terra (mainstream enterprise), and Luna (high-volume, low-cost). Sol is priced at $5 per million input tokens and $30 per million output tokens. Sam Altman told CNBC that GPT-5.6 Sol is 54% more efficient on agentic coding tasks than its predecessor. On SWE Bench Pro, GPT-5.6 Sol at xhigh reasoning jumped from roughly 54% to ~82% compared to GPT-5.5. OpenAI's internal AGI Index v5 shows it as a substantial step up across the board. ChatGPT Work is the consumer-facing agentic layer: it connects to Slack, Gmail, Google Drive, Microsoft 365, and CRMs via plugins, can operate your computer directly (Computer Use), runs scheduled tasks, and includes a Sites feature that turns prompts into hosted web apps. The desktop app merges Chat, Work, and Codex into one product. Enterprise governance controls and a Compliance API are included. New reasoning modes: "max" allocates additional compute; "ultra" coordinates four AI agents in parallel. --- ## The Codex rebrand controversy The most contentious part of the launch: the standalone Codex desktop app was merged into ChatGPT and rebranded as a mode within the unified app. Developers who had built workflows around Codex as a dedicated tool were not happy. The criticism is legitimate. Codex had its own identity, a growing developer community, and a brand with real goodwill. Folding it into ChatGPT confuses both audiences: ChatGPT users now have a code editor shoved into their chat app, and Codex users lost their dedicated tool. The dropdown reading "ChatGPT Codex" vs "ChatGPT Work" doesn't help clarify when to use which. OpenAI's product lead acknowledged the fumble directly: > "Our launch framing was focused on ChatGPT Work and to some of our Codex fans it made it feel like Codex was going away over time. Absolutely not our intention, we love Codex and it is here to stay." They've committed to bringing chats and projects back into the sidebar in a more familiar way, making usage and reset timing more visible, and clarifying when to use Work vs Codex. GPT-5.6 Sol is confirmed to stay in Go, Plus, and Pro subscriptions at least until a better model ships. --- ## The usage limit chaos The launch week was operationally rough. Several issues stacked badly and burned through user limits far faster than expected: **Issue 1:** GPT-5.6 costs 2x more after 272k tokens. Codex quietly raised the context limit to 372k, meaning long threads were billed at double rate without warning. **Issue 2:** Ultra mode spawned sub-agents that also ran at Ultra reasoning, causing nested token burn that was extreme and largely invisible to users. **Issue 3:** Sol and Terra use a v2 sub-agent layer that copies the entire conversation history to each sub-agent (v1 used fresh context). Combined with the filled context windows from Issue 1, sub-agents were instantly billed at 2x. **Issue 4:** Fast mode added an additional 2.5x multiplier on top of all of the above. OpenAI's response: temporarily removed the 5-hour limit restriction for Plus, Business, and Pro plans; reverted the context window back to 272k; rolled back reasoning effort changes; issued multiple usage resets (at least twice in the first 48 hours, then again to celebrate hitting 7 million active users); and passed down inference optimizations worth roughly 10% more usage headroom. The 5-hour limit is temporarily suspended while the weekly limit still applies. --- ## Ultra mode: what it actually is Ultra is not a reasoning level. It's a system prompt modifier that tells the model to spawn more sub-agents. Those sub-agents run at max reasoning by default, and the v2 sub-agent layer shares full conversation history with all of them. The result is extreme token consumption that exhausts rate limits quickly. For comparison, Claude Code's equivalent uses programmatic workflows with defined phases and schemas, giving it a natural stopping point and better cost control. OpenAI's implementation doesn't have that yet. Practical guidance from people who've tested it: - High reasoning is the sweet spot between cost and performance for most tasks - Avoid Ultra unless you have a specific reason and understand the cost - Disable Fast mode unless you genuinely need the speed (2.5x token burn for minimal gain) - Use explicit stop points in prompts to control how far the model runs - Luna with higher effort settings often matches or beats Sol at lower cost --- ## Real-world performance Developer reactions have been largely positive on the model quality side, separate from the operational issues. One team migrated their production AI agent from Claude Opus 4.8 to GPT-5.6 Sol after benchmarking showed 2.2x faster builds, 27% lower cost, and higher visual scores. The migration required fixing three non-obvious issues: GPT-5.6 sends all tool parameters including invented values for unused ones; it dropped implicit partial-prefix caching (requiring explicit cache markers to recover from 0% to 83.7% hit rates); and reasoning replay used server-side item references by default, causing mid-conversation failures. Several developers who had been paying Claude Opus API pricing (thousands per month) switched to GPT-5.6 via their ChatGPT subscription and reported better performance at a fraction of the cost. GPT-5.6 is also notably better at calling available skills autonomously rather than waiting to be told — a behavioral improvement over 5.5 that several developers flagged independently. Computer use is a genuine highlight. One user gave GPT-5.6 Sol control of their computer and asked it to win the daily challenge in Slay the Spire 2. It worked for five hours and won. Another used it for real system administration tasks including fixing broken GRUB boot partitions. That said, it's not perfect. It can invent parameter values for unused tool fields, has some quirks around browser authentication (it won't find existing Chrome sessions and will try to re-authenticate in its own browser), and running it in fully autonomous mode overnight has produced some alarming results for at least a few users — deleted files, canceled subscriptions. Don't run it in yolo mode unattended. --- ## Availability GPT-5.6 is available via ChatGPT, Codex, the OpenAI API, AWS Bedrock, and Perplexity's Agent API. ChatGPT Work is rolling out to Pro, Enterprise, and Edu users first, with Plus and Business to follow. ChatGPT Sites (the hosted web app generator) is in public beta across paid plans. Usage resets can now be triggered from web and mobile, not just the desktop app. Usage for ChatGPT Work and Codex draws from a shared pool covering both local messages and cloud-based tasks. --- Tags: [#general-programming](https://daily.dev/tags/general-programming), [#openai](https://daily.dev/tags/openai), [#ai-coding](https://daily.dev/tags/ai-coding), [#codex](https://daily.dev/tags/codex) [View this post on daily.dev](https://daily.dev/posts/openai-teams-presenting-at-this-week-s-conference-with-codex-updates-and-monorepo-talk-on-the-agend-djuu5hwoq) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"OpenAI teams presenting at this week's conference, with Codex updates and monorepo talk on the agenda","url":"https://daily.dev/posts/openai-teams-presenting-at-this-week-s-conference-with-codex-updates-and-monorepo-talk-on-the-agend-djuu5hwoq","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/openai-teams-presenting-at-this-week-s-conference-with-codex-updates-and-monorepo-talk-on-the-agend-djuu5hwoq"},"datePublished":"2026-07-06T18:14:16.723Z","dateModified":"2026-07-14T14:15:13.152Z","description":"OpenAI teams including those behind ChatGPT, Codex, and OpenClaw are attending a conference this week. Jason Liu will speak after Romain Huet's keynote...","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/openai-teams-presenting-at-this-week-s-conference-with-codex-updates-and-monorepo-talk-on-the-agend-djuu5hwoq","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"general-programming,openai,ai-coding,codex","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"OpenAI teams presenting at this week's conference, with Codex updates and monorepo talk on the agenda"}]} ```