The latest drop
8 September 2026
OpenAI shipped 4 image models in this window — the most recent cluster in the catalog. Of the 4, 2 have a spec page and 2 are folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 5 credits on GPT Image 2.5 Flare Edit. Highest output in it: 4K on GPT Image 2.5 Flare Edit.
| Model | Built by | Credits per job |
|---|---|---|
| GPT Image 2.5 Flare Edit | OpenAI | 5-9 credits |
| GPT Image 2.5 Flare Text to Image | OpenAI | 5-9 credits |
| GPT Image 2.5 Sunburst Edit | OpenAI | 5-9 credits |
| GPT Image 2.5 Sunburst Text to Image | OpenAI | 5-9 credits |
9 release timelines
15 launch windows from 12 providers. Busiest: 1–5 February 2026 with 19.
6 January 2026 → 25 August 2026 · 48 spec pages
13 launch windows from 20 providers. Busiest: 12–14 May 2026 with 8.
1 January 2026 → 8 September 2026 · 30 spec pages
4 launch windows from 8 providers. Busiest: 29 July – 2 September 2026 with 4.
22 January 2026 → 2 September 2026 · 13 spec pages
2 launch windows from 3 providers. Busiest: 1–24 August 2026 with 4.
6 January 2026 → 24 August 2026 · 5 spec pages
13 launch windows from 17 providers. Busiest: 4 November – 1 December 2025 with 11.
15 January 2025 → 16 December 2025 · 38 spec pages
13 launch windows from 19 providers. Busiest: 5–23 December 2025 with 7.
1 March 2025 → 23 December 2025 · 30 spec pages
2 launch windows from 5 providers. Busiest: 1 October – 16 December 2025 with 4.
2 April 2025 → 16 December 2025 · 6 spec pages
3 launch windows from 4 providers. Busiest: 1 September 2025 with 6.
1 April 2025 → 4 December 2025 · 9 spec pages
2 launch windows from 6 providers. Busiest: 15 March – 1 August 2024 with 5.
15 March 2024 → 25 November 2024 · 7 spec pages
The catalog by release year
All 330 models Versely runs, counted by the year on their release date. 323 of them sit on the 9 timelines above; everything else is listed here for completeness rather than given a page, for the reasons below.
| Year | Video | Image | Audio, lipsync & upscale |
|---|---|---|---|
| 2023 | 0 | 0 | 1 |
| 2024 | 1 | 9 | 4 |
| 2025 | 72 | 47 | 22 |
| 2026 | 93 | 61 | 20 |
What has no timeline, and why
A launch window needs at least 3 models in it to be a window rather than a row, so a year only earns a page for a content type when the catalog holds at least that many dated to it. 4 year-and-type combinations fall short — they arrive one or two at a time, months apart, which is a list rather than a release timeline. They are counted here instead of padded into pages.
| Year | Content type | Models | Short by |
|---|---|---|---|
| 2024 | Video | 1 | 2 |
| 2024 | Audio | 2 | 1 |
| 2024 | Lipsync | 2 | 1 |
| 2023 | Audio | 1 | 2 |
video upscale is refused outright — 1 model in the entire catalog, so no year can ever clear the bar. Everything on this list is still covered on the model index.
Frequently asked questions
What is the newest AI model on Versely?+
GPT Image 2.5 Sunburst Text to Image by OpenAI, released 8 September 2026. It arrived with 3 other image models in the 8 September 2026 window.
How often do new AI video models come out?+
Across the 323 video, image, audio and lipsync models these 9 timelines cover, releases arrive in bursts rather than a steady drip: 67 launch windows of 3 or more models, the largest of them carrying 19.
Which kinds of AI model get a release timeline here?+
video (2 years), image (3 years), audio (2 years) and lipsync (2 years), covering 2024 to 2026. A year only gets a page for a content type when the catalog holds at least 3 models of it dated that year, which is the same bar a single launch window has to clear. 4 combinations fell short and are counted on this page instead — 2024 video (1), 2024 audio (2), 2024 lipsync (2) and 2023 audio (1).
Why is there no timeline for AI video upscaling?+
video upscale is 1 model in the whole catalog. One model cannot form a launch window in any year, so it is counted here rather than given a page that would be a single row under a heading.
Where do these release dates come from?+
The released_at field on each of the 330 models in Versely's catalog snapshot. Nothing here is hand-dated, and the cohorts are computed from those dates at build time, so the grouping is reproducible from the same data.
Every model here, in one studio
The all-in-one AI studio for creators. 60+ models for video, image, voice, music and lipsync — in your browser or on your phone.
Free account. Works in your browser - no install needed. The same account signs in on your phone.