DEV Community

Max Bob
Max Bob

Posted on

OpenAI API deprecations 2026–2027: every shutdown date and recommended replacement (GPT-4o, GPT-4.1, GPT-5, o3, Whisper)

Short answer. GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini leave the OpenAI API on 14 January 2027. OpenAI's recommended replacement for all four is gpt-6-luna. Before that, GPT-5, GPT-5 mini, GPT-5 nano, GPT-5 pro, o3 and o3-pro shut down on 11 December 2026. After that come the legacy audio and realtime models (20 January), Whisper and the GPT-4o transcription models (26 February), and GPT-5.1 (1 April 2027).

Every date and replacement below is copied from OpenAI's deprecations page. I also compared them with the shutdown_date field that GET /v1/models returned on 27 October 2026 (UTC): for all 90 models that carry a date there, it matched the page exactly. "Days left" is counted from 28 October 2026.

Upcoming OpenAI API shutdowns

Shutdown date What shuts down Recommended replacement Notice given Days left
31 Oct 2026 Evals platform: existing evals become read-only Promptfoo migration guide 150 days 3
30 Nov 2026 Evals dashboard and API Promptfoo (same guide) 180 days 33
30 Nov 2026 Agent Builder (ChatKit stays) Agents SDK or ChatGPT Workspace Agents 180 days 33
30 Nov 2026 v1/prompts API and reusable prompt objects Move prompt text into your own code 180 days 33
1 Dec 2026 gpt-image-1-mini, gpt-image-1.5 (and -2025-12-16), chatgpt-image-latest gpt-image-2.5-sunburst or gpt-image-2.5-flare 182 days 34
11 Dec 2026 gpt-5, o3 (and their snapshots) gpt-5.6-sol 183 days 44
11 Dec 2026 gpt-5-mini gpt-5.6-terra 183 days 44
11 Dec 2026 gpt-5-nano gpt-5.6-luna 183 days 44
11 Dec 2026 gpt-5-pro, o3-pro gpt-5.6-sol with reasoning.mode: pro 183 days 44
6 Jan 2027 tts-1, tts-1-hd, gpt-4o-mini-tts (both snapshots) gpt-realtime-2.1-mini 97 days 70
6 Jan 2027 Creating new fine-tuning jobs (existing fine-tuned models keep working until their base model retires) — 244 days 70
14 Jan 2027 GPT-4o (gpt-4o-2024-08-06, gpt-4o-2024-11-20), GPT-4o mini (gpt-4o-mini-2024-07-18), GPT-4.1 (gpt-4.1-2025-04-14), GPT-4.1 mini (gpt-4.1-mini-2025-04-14) gpt-6-luna 92 days 78
20 Jan 2027 gpt-realtime, gpt-4o-realtime gpt-realtime-2.1 184 days 84
20 Jan 2027 gpt-realtime-mini, gpt-4o-mini-realtime gpt-realtime-2.1-mini 184 days 84
20 Jan 2027 gpt-audio, gpt-audio-mini, gpt-4o-audio, gpt-4o-mini-audio gpt-audio-1.5 184 days 84
20 Jan 2027 gpt-4o-mini-transcribe-2025-03-20 gpt-4o-mini-transcribe-2025-12-15 (see the catch below) 184 days 84
26 Feb 2027 whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe, gpt-4o-transcribe-diarize gpt-live-transcribe or gpt-transcribe 184 days 121
26 Feb 2027 /v1/audio/translations endpoint v1/realtime/translations 184 days 121
1 Apr 2027 gpt-5.1, gpt-5.3-codex gpt-6-sol 182 days 155
1 Apr 2027 gpt-5.4-nano gpt-6-luna 182 days 155

Fine-tuned versions of GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini follow the same 14 January schedule as their base models.

Three things the table doesn't make obvious

  1. GPT-4o got 92 days of notice, not six months. OpenAI's page says generally available models get "at least 6 months" of notice, "unless safety or compliance concerns require a faster timeline". The GPT-4 series notice went out on 14 October 2026 for a 14 January 2027 shutdown. Most other upcoming model retirements on the page got 182–184 days. If you planned around six months, you now have 78 days.
  2. One recommended replacement retires 37 days after it takes over. gpt-4o-mini-transcribe-2025-03-20 should move to gpt-4o-mini-transcribe-2025-12-15 by 20 January 2027. That model is itself shut down on 26 February 2027. Go straight to gpt-transcribe or gpt-live-transcribe and migrate once.
  3. Four model families collapse into one. GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini all map to gpt-6-luna. If you picked a mini model for cost or latency, re-run your evals and check the bill before switching. One recommendation for all four does not mean the same price or speed as each of them.

What already shut down in 2026

Shut down Models / systems Recommended replacement
23 Oct 2026 gpt-4, gpt-4-turbo, gpt-4-1106-preview, gpt-4o-2024-05-13, o1, o3-mini gpt-5.6-sol
23 Oct 2026 o1-pro gpt-5.6-sol with reasoning.mode: pro
23 Oct 2026 gpt-3.5-turbo, o4-mini gpt-5.6-terra
23 Oct 2026 gpt-4.1-nano gpt-5.6-luna
23 Oct 2026 gpt-image-1 gpt-image-2.5-sunburst or gpt-image-2.5-flare
1 Oct 2026 gpt-5.4-cyber "The most capable cyber model available to you"
28 Sep 2026 gpt-3.5-turbo-instruct, gpt-3.5-turbo-1106, babbage-002, davinci-002 gpt-5.6-terra
24 Sep 2026 Videos API and all Sora 2 models none listed
26 Aug 2026 Assistants API Responses API + Conversations API
10 Aug 2026 gpt-5.2-chat-latest, gpt-5.3-chat-latest gpt-5.6-sol
23 Jul 2026 gpt-5-chat-latest, gpt-5.1-chat-latest, gpt-5-codex, gpt-5.1-codex, gpt-5.1-codex-max, gpt-5.2-codex, o3-deep-research, o4-mini-deep-research gpt-5.6-sol
23 Jul 2026 computer-use-preview, gpt-5.1-codex-mini, gpt-4o-search-preview-2025-03-11, gpt-4o-mini-search-preview-2025-03-11 gpt-5.6-terra

Check your own account in 30 seconds

OpenAI's page says every model in GET /v1/models carries an optional shutdown_date (YYYY-MM-DD, or null if none is announced). This prints every model your key can see that has a date, and marks the ones already past it:

import datetime, json, os, urllib.request

req = urllib.request.Request(
    "https://api.openai.com/v1/models",
    headers={"Authorization": f"Bearer {os.environ['OPENAI_API_KEY']}"},
)
models = json.load(urllib.request.urlopen(req, timeout=30))["data"]
today = datetime.date.today().isoformat()

for m in sorted(models, key=lambda m: m.get("shutdown_date") or "9999"):
    d = m.get("shutdown_date")
    if d:
        print("PAST " if d < today else "     ", d, m["id"])
Enter fullscreen mode Exit fullscreen mode

Or with curl and jq:

curl -s https://api.openai.com/v1/models -H "Authorization: Bearer $OPENAI_API_KEY" \
  | jq -r '.data[] | select(.shutdown_date != null) | "\(.shutdown_date) \(.id)"' | sort
Enter fullscreen mode Exit fullscreen mode

Here is what I saw when I ran it on 27 October 2026 (UTC). The list depends on your organisation, so yours may differ.

  • My key could see 142 models. 90 had a shutdown_date, and 42 of those dates were already in the past. That included gpt-4, gpt-3.5-turbo, o1, o3-mini and o4-mini (23 October) and computer-use-preview (23 July). A model showing up in the list does not mean it still works. Compare the date with today.
  • Some snapshots have no date even though a sibling does. tts-1-1106 and tts-1-hd-1106 came back with shutdown_date: null, while tts-1 and tts-1-hd retire on 6 January 2027. The same was true of gpt-4o-search-preview (the -2025-03-11 snapshot shut down in July) and gpt-3.5-turbo-instruct-0914. The deprecations page doesn't list these IDs at all. Treat null as "unknown", not "safe".

Where to track it

Disclosure: I run llmlatency.dev, an independent tracker of AI API latency, uptime and model retirements. I have no affiliation with OpenAI. Every row in this post links back to OpenAI's own page; if you spot a mismatch, tell me in the comments and I'll fix it.

Top comments (0)