GPT-5.5 Retirement Details and Replacement Guide

OpenAI completed the GPT-5.5 retirement from ChatGPT, ChatGPT Work, and Codex on October 14, 2026 across all subscription tiers.

At Layer3Labs, we build AI workflow systems for businesses.

GPT-5.5 still runs in the OpenAI Application Programming Interface (API), and Codex users signed in with ChatGPT switch to gpt-5.6-sol.

October 14 Interface Retirement and Replacement Models

GPT-5.5 left ChatGPT, ChatGPT Work, and Codex on October 14, 2026, and it still runs in the OpenAI API. For Codex users signed in with ChatGPT, OpenAI recommends gpt-5.6-sol as the replacement model. OpenAI confirmed in the official model release notes that this removal applies to all tiers, including consumer, Business, Enterprise, and Edu plans. The model no longer appears in interface drop-down selectors for interactive chat sessions.

Codex users signed in with ChatGPT need to review saved model settings. Pinned model strings in custom agents, scheduled background jobs, and automation scripts require manual updates to gpt-5.6-sol, which the GPT-5.6 Sol vs Terra vs Luna guide compares with the other GPT-5.6 models.

• Affected interfaces: ChatGPT, ChatGPT Work, and Codex.
• Affected plans: consumer, Business, Enterprise, and Edu plans.
• Recommended Codex replacement: gpt-5.6-sol.

Model Not Found Errors and Codex Configuration Fixes

Users attempting to load retired model configurations in ChatGPT or Codex encounter an error stating that the requested model was not found. The error appears when a chat, saved setting or Codex config still names GPT-5.5. Starting a fresh conversation and choosing an available model from the current drop-down list resolves the issue in ChatGPT.

If the model selector disappears from your interface entirely, review the troubleshooting steps in the guide on the ChatGPT model picker missing. For detailed diagnostics regarding missing model errors, read the reference on ChatGPT model not found errors.

Change the model in the Codex config first, then check saved agents. Check custom scripts and scheduled tasks to replace hardcoded gpt-5.5 references with gpt-5.6-sol. Further error handling steps appear in the guide covering the Codex model not available error.

• Start a new conversation in ChatGPT and select an active model from the current list.
• Change the model in the Codex config to gpt-5.6-sol.
• Audit custom agents, scheduled tasks, and automation scripts for hardcoded references.

API Availability Following the GPT-5.5 Retirement

The retirement of GPT-5.5 from interactive interfaces does not apply to the OpenAI Application Programming Interface (API). Developers can continue running production workloads against gpt-5.5 without immediate code changes at the API tier.

Neither gpt-5.5 nor any dated GPT-5.5 snapshot appears on the official OpenAI deprecations page as of October 15, 2026, so no API shutdown date has been published. Teams can check the operational status of any model by calling the GET /v1/models endpoint or the Retrieve a model endpoint. A date in the shutdown_date field means OpenAI has scheduled that model's API shutdown.

Authentication method determines whether Codex retains access to the model. According to a press report published by 36Kr, Codex sessions authenticated with a direct API key continue to run GPT-5.5 for now. Conversely, Codex users logged in through standard ChatGPT credentials lose access and must migrate their environments.

• The OpenAI API still serves gpt-5.5 with no shutdown date listed as of October 15, 2026.
• Codex sessions authenticated with an API key keep the model for now according to press reports.
• Call GET /v1/models and read the shutdown_date field.

GPT-5.5 Retirement Timeline and Preceding Changes

GPT-5.5 completed an operational run of under six months in consumer interfaces before its October 14, 2026 retirement. Press reporting from 36Kr notes that OpenAI released the model on April 23, 2026. OpenAI announced the planned retirement on September 14, 2026 in model documentation cited by Mixed, followed by confirmation from the official ChatGPT account on September 15, 2026.

A community farewell thread for GPT-5.5 appeared on October 13, 2026, one day before the retirement, according to 36Kr.

Two earlier ChatGPT retirements moved conversations onto GPT-5.5 itself. On June 12, 2026, OpenAI retired GPT-5.2 Instant, Thinking, and Pro from ChatGPT, transferring active conversations to the matching GPT-5.5 model. Two weeks later, on June 26, 2026, GPT-4.5 left ChatGPT, and OpenAI routed existing chats to GPT-5.5.

• April 23, 2026: GPT-5.5 launches according to press reports.
• June 12, 2026: GPT-5.2 Instant, Thinking, and Pro retire from ChatGPT, with conversations moved to GPT-5.5.
• June 26, 2026: GPT-4.5 leaves ChatGPT, with conversations moved to GPT-5.5.
• September 14, 2026: OpenAI announces the October 14 interface retirement for GPT-5.5.
• October 14, 2026: GPT-5.5 retires from ChatGPT, ChatGPT Work, and Codex.

September Interface Modifications and Unpublished Policies

OpenAI changed how ChatGPT picks reasoning modes on September 14, 2026, in the same changelog entry that announced the retirement. ChatGPT stopped automatically switching Plus and Pro subscribers from Instant to Thinking modes. Users must now explicitly request deeper reasoning to engage extended processing.

OpenAI removed the Higher intelligence setting from ChatGPT on the web for Plus and Pro tiers, while Business accounts retain access under Settings > General. Free and Go users received a Think button that runs GPT-5.6 Luna. GPT-5.3-Codex-Spark was deprecated the same day.

Several practical migration questions remain unpublished in OpenAI documentation as of October 15, 2026. OpenAI has not published what happens to existing historical GPT-5.5 conversations within user histories. Furthermore, OpenAI documentation does not state whether custom GPT configurations or project workspaces pinned to GPT-5.5 switch to an alternative model automatically. Check OpenAI's model release notes for updates on both points.

• Automatic switching from Instant to Thinking mode was disabled for Plus and Pro users on September 14, 2026.
• Higher intelligence settings were removed on the web for Plus and Pro tiers but retained for Business users.
• GPT-5.3-Codex-Spark was formally deprecated on September 14, 2026.
• OpenAI has not published policies regarding existing conversations or automatic migration for custom GPT configurations.

Upcoming Model Retirement Schedules Across Major Providers

OpenAI, Google, and Anthropic each have API model shutdowns scheduled between October 22, 2026 and January 14, 2027. On October 23, 2026, OpenAI shuts down gpt-3.5-turbo, gpt-4, gpt-4-turbo, o1, o3-mini, o4-mini, and others in the API. Replacement pairings such as gpt-5.6-terra and gpt-5.6-sol are listed in the OpenAI deprecated models guide.

Two more OpenAI API shutdown dates follow. Snapshots for gpt-5, gpt-5-mini, gpt-5-nano, and o3 shut down on December 11, 2026. Five GPT-4o and GPT-4.1 snapshots (gpt-4o-2024-08-06, gpt-4o-2024-11-20, gpt-4o-mini-2024-07-18, gpt-4.1-2025-04-14 and gpt-4.1-mini-2025-04-14) retire on January 14, 2027. OpenAI announced that shutdown on October 14, 2026, the same day GPT-5.5 left ChatGPT, and named gpt-6-luna as the replacement.

Google shuts down several Gemini API preview models on October 22, 2026, including Veo 3.1 previews and Gemini Omni Flash preview, as covered in Gemini model deprecations. Anthropic retires Claude Sonnet 4.5 from the Claude API on November 30, 2026, recommending claude-sonnet-5-5 as detailed in Anthropic model deprecations.

• October 22, 2026: Gemini API preview models shut down (Veo 3.1 and Gemini Omni Flash).
• October 23, 2026: OpenAI API shuts down gpt-3.5-turbo, gpt-4, gpt-4-turbo, o1, o3-mini, o4-mini, and others.
• November 30, 2026: Anthropic retires Claude Sonnet 4.5 in favor of claude-sonnet-5-5.
• December 11, 2026: OpenAI API shuts down gpt-5, gpt-5-mini, gpt-5-nano, and o3 snapshots.
• January 14, 2027: OpenAI API retires GPT-4o and GPT-4.1 snapshots in favor of gpt-6-luna.

Action Steps Following the GPT-5.5 Retirement

Search repositories and Codex configs for pinned gpt-5.5 references. Change the ones that run through Codex with a ChatGPT sign-in to gpt-5.6-sol, since calls made with an OpenAI API key can stay on gpt-5.5 for now.

Organizations that rely on specific response characteristics should evaluate gpt-5.6-sol in staging environments before rolling out script updates. If your systems require continued access to GPT-5.5, transition affected pipelines to the OpenAI API using valid API keys. Keep model names in one config file so the next retirement is a one-line change.

In ChatGPT itself there is no setting to change: start a new chat and pick a model the selector shows, or leave the selector on Auto. If the error keeps appearing after that, the ChatGPT model not found guide lists the ordered checks, from signing out and back in to contacting OpenAI support with the exact error message.