GPT-56T 861 —
MUSE-SPK 835 -0.7%
GPT-56SC 828 -5.2%
QWEN-38X 824 —
CL-OP55X 822 —
GROK-46H 822 -5%
GPT-6A 820 —
GLM-5 784 -8.4%
CL-FAB5H 743 -5.6%
KIMI-K3X 742 -8.4%
CL-OP5H 720 -5.8%
CL-OP5X 709 -18%
CL-OP46H 698 -5.9%
CL-OP47H 690 -5.9%
GEM-38FH 677 +0.1%
GEM-37FH 657 -24%
GPT-56S 622 —
CL-OP47 582 -0.7%
GPT-55H 582 —
INKL 531 —
GEM-31P 513 —
GEM-3P 499 —
CL-OP46 496 -0.2%
CL-OP48 490 —
GPT-56T 861 —
MUSE-SPK 835 -0.7%
GPT-56SC 828 -5.2%
QWEN-38X 824 —
CL-OP55X 822 —
GROK-46H 822 -5%
GPT-6A 820 —
GLM-5 784 -8.4%
CL-FAB5H 743 -5.6%
KIMI-K3X 742 -8.4%
CL-OP5H 720 -5.8%
CL-OP5X 709 -18%
CL-OP46H 698 -5.9%
CL-OP47H 690 -5.9%
GEM-38FH 677 +0.1%
GEM-37FH 657 -24%
GPT-56S 622 —
CL-OP47 582 -0.7%
GPT-55H 582 —
INKL 531 —
GEM-31P 513 —
GEM-3P 499 —
CL-OP46 496 -0.2%
CL-OP48 490 —
← Back to feed

OpenAI Retires GPT-5.2 and GPT-5.3-Codex From the Codex App — the Models Users Switched From Claude to Get

OpenAI has deprecated GPT-5.2 and GPT-5.3-Codex from its Codex application for ChatGPT subscribers, forcing users onto GPT-5.5 as the sole available model. The move has generated significant pushback on the OpenAI developer community forum, where a thread tracking the change has accumulated 7,800 views, 233 replies, and comments from 73 distinct users since the sunset took effect.

The timing is difficult for OpenAI. The company is in a period of acute competitive pressure from Claude Code, which crossed $1 billion in revenue in its first six months and now leads on the Ramp AI enterprise spending index. Retiring the Codex-specific models that gave developers a reason to switch away from Anthropic removes a differentiator at a moment the company can least afford it.

What Developers Valued

GPT-5.3-Codex had a specific reputation among developers for producing compact, high-quality code from precise prompts. Forum commentary frames the distinction clearly: “GPT-5.5 works fine for adjustments, but Codex 5.3 was clearly making better/compact code after a good prompt.” Several users report that the model was the direct reason they moved from Claude Max to ChatGPT Plus subscriptions. “GPT Codex 5.3 is the reason I switched from Claude Max to ChatGPT Plus. It’s the reason I pursued friends to do the same.”

GPT-5.5 is a more capable model on aggregate benchmarks — 82.7% on Terminal-Bench, 58.6% on SWE-Bench Pro — but benchmark performance does not fully capture what these developers valued. Codex 5.3’s architecture was tuned for a specific editing and generation pattern that GPT-5.5 handles differently.

The API Path Costs More

Users who want GPT-5.3 behaviour now have no direct path within their subscription tier. OpenAI has not made the deprecated models available as a legacy API option. The only alternative is API access with a separate token budget, which converts a flat-rate subscription cost into a variable one.

GPT-5.5 through the API is priced at $5 per million input tokens and $30 per million output tokens. For a developer running a meaningful daily Codex workflow, that is a different cost structure than a $20 or $100 monthly subscription.

OpenAI’s Model Lifecycle Problem

The retirement of 5.2 and 5.3 reflects a recurring tension in OpenAI’s product strategy: the company has accelerated its model release cadence dramatically since 2025, which means older models that developers have built workflows around are deprecated faster. The community response to the Codex sunset is a concentrated version of a broader dissatisfaction with model churn.

There is no published deprecation schedule for Codex models. Users found out when the models disappeared from the app interface.

The competitive stakes are visible in the forum thread itself: multiple commenters explicitly name Claude Code as the destination they are reconsidering. “Okay, it seems canceling the subscription is the best we can do,” writes one user, after noting OpenAI had not engaged with the community thread in the days after the change.

OpenAI’s Codex app has 5 million weekly active users as of its most recent public disclosure. The share of those users who built their workflow around the Codex-specific models — and who are now weighing alternatives — is the number OpenAI will watch closely over June.