OpenAI retires additional legacy GPT and reasoning model snapshots

What changed

In the April 22, 2026 deprecation entry, OpenAI announced shutdown of a set of older model snapshots on October 23, 2026, with recommended substitutes toward the GPT-5.6 family (and gpt-image-2 for gpt-image-1).

Examples from the official table include:

• gpt-4o-2024-05-13 → gpt-5.6-sol
• gpt-4-0613 / related GPT-4 snapshots → gpt-5.6-sol
• gpt-4.1-nano / gpt-4.1-nano-2025-04-14 → gpt-5.6-luna
• o1-2024-12-17, o1-pro-2025-03-19, o3-mini-2025-01-31, o4-mini-2025-04-16 → GPT-5.6 substitutes as listed
• gpt-image-1 → gpt-image-2
• Multiple fine-tuned bases (ft-gpt-3.5-turbo, ft-gpt-4, ft-gpt-4.1-nano-2025-04-14, and others) with listed replacement base models

Consult the full table on OpenAI’s deprecations page for every snapshot ID in this batch.

Who is affected

Applications, eval suites, and fine-tuned deployments still pinned to any snapshot in the October 23, 2026 shutdown list—especially long-lived aliases such as older gpt-4o and GPT-4 snapshots.

What you need to do

Required before October 23, 2026 for any listed snapshot you still call:

1. Export usage by model ID and map each to OpenAI’s recommended substitute.
2. Update model strings (and fine-tune bases where applicable).
3. Re-run quality, latency, and safety checks against replacements before cutover.
4. Remove remaining calls to retiring snapshots before the shutdown date.

Migration effort

Medium is an editorial estimate for a typical model-ID swap plus regression testing—not a claim that every snapshot in this batch has the same complexity. Effort varies by model, whether you use fine-tuned derivatives of retiring bases, and how tightly behavior was pinned to a specific snapshot. Some integrations may be lower; fine-tune or deeply customized stacks may be higher.

Cost impact

Unknown. Replacements are in a newer model family with different pricing; OpenAI’s deprecation notice does not quantify spend change for equivalent workloads. Compare current vs replacement pricing against your token mix.