GPT
OpenAI · 18 tracked releases · March 14, 2023 – September 29, 2026
At a glance
| Releases tracked | 18 |
|---|---|
| First tracked | March 14, 2023 — GPT-4 |
| Latest | September 29, 2026 — GPT-6.1 Sol |
| Span | 1,295 days |
| Median gap | 29 days between releases |
| Weights | Proprietary throughout |
| Record quality | 8 of 18 verified · how this is judged |
Release cadence
One tile per tracked release, placed in the month it shipped, so the pauses in this line are the pauses in the data. Years with nothing tracked keep their row.
- 20231
- 20241
- 2025
- GPT-4.5, OpenAI, February 27, 2025
- GPT-5, OpenAI, August 7, 2025
- GPT-5.1, OpenAI, November 12, 2025
- GPT-5.2, OpenAI, December 11, 2025
- 2026
- GPT-5.3-Codex, OpenAI, February 5, 2026
- GPT-5.4, OpenAI, March 5, 2026
- GPT-5.5, OpenAI, April 23, 2026
- GPT-5.6 Sol, OpenAI, July 9, 2026GPT-5.6 Luna, OpenAI, July 9, 2026GPT-5.6 Terra, OpenAI, July 9, 2026
- GPT-5.6-Cyber, OpenAI, August 10, 2026
- GPT-6 Astra, OpenAI, September 3, 2026GPT-Rosalind, OpenAI, September 8, 2026GPT-6 Sol, OpenAI, September 22, 2026GPT-6 Luna, OpenAI, September 22, 2026GPT-6.1 Sol, OpenAI, September 29, 2026
Lineage
Ordered by announcement date, and spaced by the gap between releases, so a quiet year looks like one. Lineage is not inferred from version numbers — labs do not number consistently, so the dates decide the order.
-
GPT-4
vision
-
+426 days
GPT-4oaudio
-
+290 days
GPT-4.5from GPT-4 Context window 8K → 128K
-
+161 days
GPT-5reasoningcodingstructured outputfunction calling
from GPT-4.5 Context window 128K → 400Kfirst evidenced: reasoning, coding, structured output, function calling
-
+97 days
-
+29 days
-
+56 days
GPT-5.3-Codexagentic
-
+28 days
GPT-5.4from GPT-5.2 Context window 400K → 1.05Mfirst evidenced: vision
-
+49 days
GPT-5.5from GPT-5.4 first evidenced: agentic
-
+32 days
-
+24 days
-
+5 days
-
+7 days
GPT-6.1 Solfrom GPT-6 Sol first evidenced: coding, structured output, function callingfirst evidenced: image in
Context window over time
Only releases with a disclosed context window appear — 2 of 18 are not shown.
What changed
Generation to generation, from evidenced values only. A field is compared only where both releases record it — otherwise it is called out as not comparable, so a research gap never reads as a change the lab made.
Across every generation of this family, parameters could not be compared — the values are not recorded on both sides of any step. That is a gap in this dataset, not a statement about the models. See data quality.
GPT-4.5 → GPT-5 161 days
- Context window
- 128K → 400K
- Capabilities first evidenced
- + reasoning + coding + structured output + function calling
GPT-5.6 Sol → GPT-6 Sol 75 days
No change in any field both releases record.
Not comparable here: context window (not researched on both releases); modalities (GPT-6 Sol evidences fewer than GPT-5.6 Sol, which means not evidenced rather than removed).
GPT-5.6 Luna → GPT-6 Luna 75 days
No change in any field both releases record.
GPT-6 Sol → GPT-6.1 Sol 7 days
- Capabilities first evidenced
- + coding + structured output + function calling
- Modalities first evidenced
- + image in
Not comparable here: context window (not researched on both releases).
Compare the two most recent side by side →
All releases from OpenAI →