--- type: "post" id: "openai-retires-gpt-5-5-from-chatgpt-chatgpt-f7ea00" slug: "openai-retires-gpt-5-5-from-chatgpt-chatgpt-f7ea00" title: "OpenAI retires GPT-5.5 from ChatGPT, ChatGPT Work and Codex; API access unaffected" description: "OpenAI’s release notes record two distinct changes: GPT-5.5 was retired from ChatGPT, ChatGPT Work and Codex on October 14, while API access was unaffected; GPT-6.1 Terra began rolling out on October 26 to the API, Work and Codex." retrieval_nugget: "OpenAI’s release notes record two distinct changes: GPT-5.5 was retired from ChatGPT, ChatGPT Work and Codex on October 14, while API access was unaffected; GPT-6.1 Terra began rolling out on October 26 to the API, Work and Codex. For builders, the distinction between product access and API availability matters for migration planning and model costs." published_at: "2026-11-01" updated_at: "2026-11-01" record_date: "2026-11-01" date_kind: "published_at" topics: ["codex"] entities: ["OpenAI","GPT-5.5","GPT-6.1 Terra","ChatGPT","Codex"] related_patterns: ["harness-architecture-outlives-model-choice"] novelty_label: "new_fact" what_changed: "The retirement now covers ChatGPT and ChatGPT Work as well as Codex; API access is unaffected." prior_refs: [] source_url: "https://help.openai.com/en/articles/9624314-model-release-notes" source_urls: ["https://help.openai.com/en/articles/9624314-model-release-notes"] schema_version: "newruntime-agent-readable-v0.2" stable_id: "post:openai-retires-gpt-5-5-from-chatgpt-chatgpt-f7ea00" status: "published" visuals: [] editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]} routes: {"html":"https://newruntime.com/posts/openai-retires-gpt-5-5-from-chatgpt-chatgpt-f7ea00/","markdown":"https://newruntime.com/posts/openai-retires-gpt-5-5-from-chatgpt-chatgpt-f7ea00.md","json":"https://newruntime.com/posts/openai-retires-gpt-5-5-from-chatgpt-chatgpt-f7ea00.json"} markdown_url: "https://newruntime.com/posts/openai-retires-gpt-5-5-from-chatgpt-chatgpt-f7ea00.md" --- # OpenAI retires GPT-5.5 from ChatGPT, ChatGPT Work and Codex; API access unaffected ## Retrieval answer OpenAI’s release notes record two distinct changes: GPT-5.5 was retired from ChatGPT, ChatGPT Work and Codex on October 14, while API access was unaffected; GPT-6.1 Terra began rolling out on October 26 to the API, Work and Codex. For builders, the distinction between product access and API availability matters for migration planning and model costs. ## Summary OpenAI’s release notes record two distinct changes: GPT-5.5 was retired from ChatGPT, ChatGPT Work and Codex on October 14, while API access was unaffected; GPT-6.1 Terra began rolling out on October 26 to the API, Work and Codex. For builders, the distinction between product access and API availability matters for migration planning and model costs. ## What happened OpenAI’s Model Release Notes say GPT-5.5 was retired from ChatGPT, ChatGPT Work and Codex on October 14, 2026. The notice explicitly says the change does not affect API access. It does not describe an API sunset or give a date for one. Teams using GPT-5.5 in the ChatGPT interface, Work or Codex therefore face a product-level model change; the notes do not say that API users must migrate. A separate entry, dated October 26, says GPT-6.1 Terra is rolling out in the OpenAI API under the identifier `gpt-6.1-terra`, as well as in ChatGPT Work and Codex. OpenAI lists visual reasoning, computer use and writing among the model’s task areas. The notes describe a rollout, not universal availability on a fixed date, and direct readers to the model directory for API capabilities and availability. They do not state that Terra is a replacement for GPT-5.5 or prescribe a migration path between the two models. The stated standard API rates are $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. API use is billed separately from ChatGPT subscriptions. These figures are the source’s standard rates; the notes point to separate credit-based and Enterprise token-based rate cards for other pricing details. In ChatGPT, Terra is available in Work and Codex, but not the Chat tab. OpenAI also says Terra does not support UAE inference residency. The restriction is based on where inference runs, not on the user’s location. The notes do not give a timetable for adding that support or further specify the geographic infrastructure involved. ## What the archive adds This is a new fact on a known OpenAI model-and-product story, rather than a wholly new thread. [OpenAI to merge ChatGPT and Codex usage into one token-based allowance for Plus, Pro and Business in December](https://newruntime.com/posts/openai-to-merge-chatgpt-and-codex-usage-into-081c83/) covered shared product allowances; the current release notes separately clarify GPT-5.5’s retirement scope and Terra’s availability and API rates. Earlier channel coverage also reported GPT-5.5’s planned retirement from ChatGPT, Work and Codex on October 14; the present notice confirms that it took effect and that API access is unaffected. ## Trend signal This supports [Harness architecture outlives model choice](https://newruntime.com/patterns/harness-architecture-outlives-model-choice/): access to the same model can differ across an API and product surfaces, while another model is introduced across selected tools. It is a concrete reminder that teams should track the model, interface and runtime path separately rather than treating a model name as a single uniform dependency. ## Why now The release notes date the GPT-5.5 retirement to October 14 and Terra’s rollout to October 26, 2026. They establish the sequence but do not explain why OpenAI chose these dates or whether Terra is intended as a direct replacement for GPT-5.5. ## Connections to other events OpenAI’s earlier release notes, as reported in the channel, announced the same October 14 retirement. A separate Field Note on shared ChatGPT and Codex allowances describes how product-level usage is being organized; this notice adds a distinct model-availability change and API pricing, not a change to those allowances. ## Why it matters A model retirement is not necessarily a single migration event. OpenAI’s notice draws a boundary between ChatGPT, Work and Codex on one side and API access on the other. A team that relies on the API should not infer a forced migration from a product retirement notice; teams using the named product surfaces should verify which model now handles their workflows and test alternatives against their own tasks. Terra’s separate availability and price details also make budgeting and deployment constraints visible. Its input and output rates differ, cached input has a lower stated price, and API billing is separate from ChatGPT subscriptions. Its absence from the Chat tab and lack of UAE inference residency may matter to product choice or deployment requirements. These facts support a targeted inventory, not an assumption that Terra is a drop-in replacement or that it will suit every workload. ## What happens next - **GPT-5.5 API access will remain available beyond the next three months.** (3 months, 65%). Watch for: OpenAI publishing an API deprecation notice or continuing to list the model in its API model directory. ## Limits and uncertainty The release notes establish the retirement scope and Terra’s rollout surfaces, listed tasks, standard API rates and UAE residency limitation. They do not provide comparative performance results, migration instructions, a deadline for Terra’s rollout to finish, or a claim that Terra replaces GPT-5.5. The stated prices do not establish total task costs, which depend on usage and the mix of input, cached input and output tokens. Nor does the source explain whether or when UAE inference residency will become available. ## Skill to build now Inventory GPT-5.5 dependencies by surface: ChatGPT, Work, Codex or API. For each workflow, record the model identifier, whether it depends on UAE inference residency, and a representative test task. This separates workflows affected by the retirement from those whose API access is explicitly unaffected, and gives a concrete basis for evaluating Terra without assuming equivalent behavior. ## Reader takeaway GPT-5.5’s retirement applies to ChatGPT, Work and Codex, not the API, according to OpenAI’s notes. GPT-6.1 Terra is rolling out across the API, Work and Codex with published token prices, but not in ChatGPT’s Chat tab and not with UAE inference residency. Check the actual surface and deployment needs before planning a migration. Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json ## Continue in the graph Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`. - topic hub: [Codex](https://newruntime.com/topics/codex.md): Explore the codex topic hub. - collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis. - collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites. - whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)