The 2025 image timeline
1–31 March 2025
3 models · 3 providersHiggsField, Reve and Runway shipped 3 image models between 1 March 2025 and 31 March 2025. All 3 have a spec page. Cheapest complete job in the window: 4 credits on Reve Text to Image. Highest output in it: 4K on Higgsfield.
- New names on the roster: HiggsField, Reve, Runway.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| HiggsfieldHiggsField's premium still generator — high-drama, cinematic frames with pronounced lighting and motion energy. The high-impact… | HiggsField | 19 credits | 4K | 1:1, 16:9, 9:16, 2:3, 3:2 |
| Reve Text to ImageReve's image model — tight prompt adherence and clean, legible text rendering. Delivers composed, poster-ready frames from dense… | Reve | 4 credits | 4K | 16:9, 3:2, 4:3, 1:1, 3:4, 2:3 |
| Runway Gen4 ImageRunway Gen4 image generation model with high-quality creative outputs and style control | Runway | 4-7 credits | 1080p | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 |
1–8 April 2025
3 models · 3 providersMidjourney, PrunaAI and Runway shipped 3 image models between 1 April 2025 and 8 April 2025. Of the 3, 1 has a spec page and 2 are folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 1 credit on Pruna Text to Image. Highest output in it: 4K on Pruna Text to Image.
- New name on the roster: PrunaAI.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| Pruna Text to ImagevariantPrunaAI's speed-optimised text-to-image — a compressed model tuned for throughput, generating usable frames at the lowest credit… | PrunaAI | 1 credit | 4K | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2 |
| Midjourney V7Midjourney V7 text-to-image generation with photorealistic and artistic outputs | Midjourney | 8 credits | — | 1:1, 16:9, 9:16, 4:3, 3:4, 2:3 |
| Runway Gen4 Image TurbovariantFast variant of Runway Gen4 image generation for rapid creative iteration with quality outputs | Runway | 3 credits | 1080p | 16:9, 9:16, 4:3, 3:4, 1:1, 21:9 |
23 April – 29 May 2025
3 models · 2 providersFlux and OpenAI shipped 3 image models between 23 April 2025 and 29 May 2025. Of the 3, 2 have a spec page and 1 is folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 1 credit on Flux Kontext. Highest output in it: 4K on GPT Image 1 Edit.
- New name on the roster: OpenAI.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| GPT Image 1 EditInstruction-based image editing with OpenAI's GPT Image — describe the change in plain English and it repaints only what you… | OpenAI | 2 credits | 4K | 1:1, 2:3, 3:2, 16:9, 9:16 |
| Flux KontextContext-aware generation with coherent compositions | Flux | 1 credit | 4K | 1:1, 9:16, 16:9, 2:3, 3:2, 3:4 |
| Flux Kontext MaxvariantMaximum context-aware image generation | Flux | 7 credits | 4K | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2 |
1 June – 28 July 2025
4 models · 3 providersGrok, SeedVR and Topaz Labs shipped 4 image models between 1 June 2025 and 28 July 2025. All 4 have a spec page. Cheapest complete job in the window: 1 credit on SeedVR Upscale. Highest output in it: 8K on SeedVR Upscale.
- New names on the roster: SeedVR, Grok.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| Topaz Image ColorizationAI-powered colorization of black-and-white images with natural, realistic color restoration | Topaz Labs | 7 credits | — | — |
| SeedVR UpscaleSeedVR from ByteDance upscales a photo with a diffusion restoration model that adds fine detail. Billed per megapixel; save it as… | SeedVR | 1 credit | 8K | 1:1, 2:3, 3:2, 16:9, 9:16 |
| Grok Imagine ImagexAI's Grok Imagine for still images — fast, loosely-styled generations with a distinctive creative bent. A strong ideation model… | Grok | 2 credits | 4K | 20:9, 19.5:9, 2:1, 16:9, 3:2, 4:3 |
| Grok Imagine Image QualityxAI Grok Imagine Image (Quality tier) text-to-image. Quality-optimized generation up to 2K resolution with a wide range of aspect… | Grok | 4 credits | 2K | 20:9, 19.5:9, 2:1, 16:9, 3:2, 4:3 |
4–5 August 2025
3 models · 2 providersLeonardo and Qwen shipped 3 image models between 4 August 2025 and 5 August 2025. Of the 3, 2 have a spec page and 1 is folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 1 credit on Qwen Image. Highest output in it: 4K on Qwen Image.
- New names on the roster: Qwen, Leonardo.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| Qwen ImageHigh-quality realistic images with excellent detail | Qwen | 1 credit | 4K | 16:9, 3:2, 4:3, 1:1, 3:4, 2:3 |
| Qwen Image MaxvariantMaximum quality image generation with Qwen's most advanced visual AI model | Qwen | 6 credits | — | 1:1, 16:9, 9:16, 4:3, 3:4 |
| Leonardo Lucid OriginLeonardo's Lucid Origin — a versatile image model spanning photoreal to stylised output with strong adherence to detailed art… | Leonardo | 4 credits | 4K | 1:1, 16:9, 9:16, 3:2, 2:3, 4:5 |
19–26 August 2025
3 models · 2 providersGoogle and Qwen shipped 3 image models between 19 August 2025 and 26 August 2025. Of the 3, 2 have a spec page and 1 is folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 1 credit on Qwen Image Edit. Highest output in it: 4K on Qwen Image Edit.
- Raised the reference-image ceiling to 14 — Nano Banana Edit.
- New name on the roster: Google.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| Qwen Image EditAlibaba Qwen's image editor — precise instruction-based edits with particular strength at changing or correcting text inside an… | Qwen | 1 credit | 4K | — |
| Nano BananaFast and efficient image generation with vibrant colors | 2 credits | 4K | 1:1, 9:16, 16:9, 3:4, 4:3, 3:2 | |
| Nano Banana EditvariantImage editing capabilities with NANO Banana model | 2 credits | 4K | 21:9, 16:9, 3:2, 4:3, 5:4, 1:1 |
9 September 2025
4 models · 1 providerByteDance shipped 4 image models on 9 September 2025. Of the 4, 3 have a spec page and 1 is folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 1 credit on Seedream 4.0. Highest output in it: 4K on ByteDance Seedream 4.
- New name on the roster: ByteDance.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| ByteDance Seedream 4Seedream 4 — the previous-generation ByteDance image model. Still a capable, lower-cost text-to-image pick when you don't need… | ByteDance | 3 credits | 4K | 1:1, 4:3, 3:4, 16:9, 9:16, 3:2 |
| Seedream 4.0ByteDance Seedream 4.0 text-to-image. High-resolution generation (up to 4K) with multi-image output and built-in prompt… | ByteDance | 1 credit | 4K | 21:9, 16:9, 3:2, 4:3, 1:1, 3:4 |
| Seedream 4.0 EditvariantImage editing with Seedream 4.5 | ByteDance | 1 credit | 4K | 21:9, 16:9, 3:2, 4:3, 1:1, 3:4 |
| Seedream EditByteDance's Seedream editor — instruction-driven changes to an existing image: replace objects, adjust backgrounds, fix details.… | ByteDance | 1 credit | 4K | 1:1, 2:3, 3:2, 16:9, 9:16 |
22–24 September 2025
3 models · 2 providersQwen and Wan shipped 3 image models between 22 September 2025 and 24 September 2025. Of the 3, 1 has a spec page and 2 are folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 3 credits on Qwen Image Edit 2509. Highest output in it: 4K on Wan 2.5 Image to Image.
- New name on the roster: Wan.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| Qwen Image Edit 2509variantQwen Image Edit 2509 is an image editing model from Alibaba that enables precise modifications to existing images based on… | Qwen | 3 credits | — | 1:1, 16:9, 9:16, 4:3, 3:4 |
| Wan 2.5 Image to ImagevariantWan 2.5's image-to-image mode — restyle, rework or upgrade an existing image while preserving its structure. Convert drafts… | Wan | 4 credits | 4K | 16:9, 4:3, 1:1, 3:4, 9:16 |
| Wan Preview 2.5 Text to ImageWan 2.5 in still-image mode — the video family's frame generator used standalone. Produces cinematic, video-native compositions… | Wan | 4 credits | 4K | 1:1, 4:3, 3:4, 16:9, 9:16 |
1–20 November 2025
3 models · 2 providersGoogle and Imagine shipped 3 image models between 1 November 2025 and 20 November 2025. Of the 3, 1 has a spec page and 2 are folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 4 credits on ImagineArt 1.5. Highest output in it: 4K on ImagineArt 1.5.
- New name on the roster: Imagine.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| ImagineArt 1.5Imagine Art 1.5 — stylised, mood-led image generation that leans expressive rather than literal. Suits emotive social visuals,… | Imagine | 4 credits | 4K | 3:1, 16:9, 3:2, 4:3, 1:1, 3:4 |
| Nano Banana ProvariantProfessional image generation with Nano Banana Pro | 6-12 credits | 4K | 1:1, 2:3, 3:2, 4:3, 3:4, 16:9 | |
| Nano Banana Pro EditvariantProfessional image editing with Nano Banana Pro | 5-8 credits | 4K | 21:9, 16:9, 3:2, 4:3, 5:4, 1:1 |
25 November 2025
3 models · 1 providerFlux shipped 3 image models on 25 November 2025. Of the 3, 1 has a spec page and 2 are folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 3 credits on Flux 2 Pro.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| Flux 2 FlexFlexible image generation and editing model from the Flux 2 family with versatile creative controls | Flux | 4 credits | — | 16:9, 9:16, 1:1, 4:3, 3:4 |
| Flux 2 ProvariantProfessional-grade image generation with the Flux 2 Pro model for high-quality detailed outputs | Flux | 3 credits | — | 16:9, 9:16, 1:1, 4:3, 3:4 |
| Flux 2 Pro EditvariantFlux 2 Pro is an advanced image editing model that allows users to modify and enhance images based on specific instructions. It… | Flux | 3 credits | — | 16:9, 4:3, 1:1, 3:4, 9:16, auto |
27 November – 1 December 2025
3 models · 3 providersKling, Qwen and Z-Image shipped 3 image models between 27 November 2025 and 1 December 2025. All 3 have a spec page. Cheapest complete job in the window: 1 credit on Z Image Turbo. Highest output in it: 4K on Z Image Turbo.
- New names on the roster: Z-Image, Kling.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| Qwen Z ImageAdvanced image generation model with enhanced creative capabilities | Qwen | 2 credits | — | 21:9, 16:9, 3:2, 4:3, 1:1, 3:4 |
| Z Image TurboZ-Image Turbo — an efficiency-tuned text-to-image model built for speed. Single-credit generations that hold up remarkably well… | Z-Image | 1 credit | 4K | 1:1, 16:9, 9:16, 4:3, 3:4 |
| Kling Image O1Kling Image O1 is an advanced image generation model by Kuaishou that produces high-quality, detailed images from text prompts.… | Kling | 3 credits | 2K | 21:9, 16:9, 3:2, 4:3, 1:1, 3:4 |
4 December 2025
3 models · 1 providerByteDance shipped 3 image models on 4 December 2025. Of the 3, 1 has a spec page and 2 are folded into a parent model's page as tier or mode variants. Cheapest complete job in the window: 4 credits on Seedream 4.5. Highest output in it: 4K on Seedream 4.5.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| Seedream 4.5ByteDance's Seedream 4.5 all-round image model — generation, editing and image-to-image in one endpoint, with the strong text… | ByteDance | 4 credits | 4K | 1:1, 2:3, 3:2, 4:3, 3:4, 16:9 |
| Seedream 4.5 EditvariantSeedream 4.5 in edit mode — upload an image and describe the revision. Handles object swaps, background changes and text… | ByteDance | 4 credits | 4K | 1:1, 2:3, 3:2, 4:3, 3:4, 16:9 |
| Seedream 4.5 Text to ImagevariantSeedream 4.5 tuned for pure text-to-image — detailed scene construction from long prompts, reliable typography, and consistent… | ByteDance | 4 credits | 4K | 1:1, 4:3, 3:4, 16:9, 9:16 |
5–23 December 2025
7 models · 5 providersFlux, Meituan and OpenAI and 2 other providers shipped 7 image models between 5 December 2025 and 23 December 2025. Of the 7, 5 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: 1 credit on Wan 2.6 Text to Image. Highest output in it: 4K on GPT Image 1.5.
- Raised the reference-image ceiling to 16 — GPT Image 1.5.
- New name on the roster: Meituan.
| Model | Built by | Credits per job | Max output | Aspect ratios |
|---|---|---|---|---|
| LongCat EditLongcat Edit is an image editing model that supports long-context image understanding and precise editing based on detailed… | Meituan | 12 credits | — | — |
| Flux 2 MaxMaximum quality image generation with the Flux 2 Max model for ultra-detailed photorealistic outputs | Flux | 6 credits | — | 16:9, 9:16, 1:1, 4:3, 3:4 |
| Flux 2 Max EditvariantMaximum quality image generation with the Flux 2 Max model for ultra-detailed photorealistic outputs | Flux | 6 credits | — | 16:9, 4:3, 1:1, 3:4, 9:16, auto |
| GPT Image 1.5OpenAI's GPT Image 1.5 — conversational prompt understanding across text-to-image, image-to-image and edits. Strong at following… | OpenAI | 2 credits | 4K | 1:1, 2:3, 3:2 |
| Wan 2.6 ImagevariantImage-to-image transformation with WAN V2.6 | Wan | 3 credits | 4K | 16:9, 4:3, 1:1, 3:4, 9:16 |
| Wan 2.6 Text to ImageText-to-image generation with WAN V2.6 | Wan | 1 credit | 4K | 1:1, 4:3, 3:2 |
| Qwen Image Edit 2511Qwen Image Edit 2511 is an image editing model from Alibaba that enables precise modifications to existing images based on… | Qwen | 1 credit | — | 1:1, 5:4, 4:5, 5:6, 3:2 |
New AI image models by month in 2025
9 of the twelve months carried a image release; the busiest window was 5–23 December 2025, with 7. Each month links to the launch window where it starts.
Who shipped image in 2025
ByteDance (7), Flux (7) and Qwen (6) led on volume. SKUs, not quality — four tiers of one model count four times.
| Provider | Models | Spec pages | First | Latest |
|---|---|---|---|---|
| ByteDance | 7 | 4 | 9 September 2025 | 4 December 2025 |
| Flux | 7 | 3 | 29 May 2025 | 16 December 2025 |
| Qwen | 6 | 4 | 4 August 2025 | 23 December 2025 |
| 4 | 1 | 26 August 2025 | 20 November 2025 | |
| Wan | 4 | 2 | 24 September 2025 | 16 December 2025 |
| Grok | 2 | 2 | 28 July 2025 | 28 July 2025 |
| OpenAI | 2 | 2 | 23 April 2025 | 16 December 2025 |
| Runway | 2 | 1 | 31 March 2025 | 8 April 2025 |
| HiggsField | 1 | 1 | 1 March 2025 | 1 March 2025 |
| Imagine | 1 | 1 | 1 November 2025 | 1 November 2025 |
| Kling | 1 | 1 | 1 December 2025 | 1 December 2025 |
| Leonardo | 1 | 1 | 5 August 2025 | 5 August 2025 |
| Meituan | 1 | 1 | 5 December 2025 | 5 December 2025 |
| Midjourney | 1 | 1 | 3 April 2025 | 3 April 2025 |
| PrunaAI | 1 | 1 | 1 April 2025 | 1 April 2025 |
| Reve | 1 | 1 | 25 March 2025 | 25 March 2025 |
| SeedVR | 1 | 1 | 11 June 2025 | 11 June 2025 |
| Topaz Labs | 1 | 1 | 1 June 2025 | 1 June 2025 |
| Z-Image | 1 | 1 | 27 November 2025 | 27 November 2025 |
What 2025 moved
Firsts, measured against every image model released before them. “New name on the roster” is the provider label in the catalog, not the company — one lab can hold several labels.
- 1–31 March 2025
- New names on the roster: HiggsField, Reve, Runway.
- 1–8 April 2025
- New name on the roster: PrunaAI.
- 23 April – 29 May 2025
- New name on the roster: OpenAI.
- 1 June – 28 July 2025
- New names on the roster: SeedVR, Grok.
- 4–5 August 2025
- New names on the roster: Qwen, Leonardo.
- 19–26 August 2025
- Raised the reference-image ceiling to 14 — Nano Banana Edit.
- New name on the roster: Google.
- 9 September 2025
- New name on the roster: ByteDance.
- 22–24 September 2025
- New name on the roster: Wan.
- 1–20 November 2025
- New name on the roster: Imagine.
- 27 November – 1 December 2025
- New names on the roster: Z-Image, Kling.
- 5–23 December 2025
- Raised the reference-image ceiling to 16 — GPT Image 1.5.
- New name on the roster: Meituan.
Dates are each model’s released_at value, walked in order and grouped until a window held 3 or more. Credits are what one complete generation costs per the model’s own price matrix; where it states none, the headline rate is shown and the model sits out the cheapest-in-window line.
Other release years
Frequently asked questions
How many AI image models were released in 2025?+
Versely's catalog carries 45 image models with a 2025 release date, from 19 providers, arriving in 13 launch windows between 1 March 2025 and 23 December 2025.
What was the biggest AI image launch of 2025?+
5–23 December 2025, with 7 image models from Flux, Meituan, OpenAI, Qwen and Wan. Of the 7, 5 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: 1 credit on Wan 2.6 Text to Image. Highest output in it: 4K on GPT Image 1.5.
Which company released the most AI image models in 2025?+
ByteDance, with 7 of the 45 image models dated 2025 — first on 9 September 2025, most recently on 4 December 2025. Flux shipped 7, Qwen shipped 6, Google shipped 4.
What changed in AI image generation in 2025?+
Measured against everything the catalog carried before it: New names on the roster: HiggsField, Reve, Runway; New name on the roster: PrunaAI; New name on the roster: OpenAI.
Run any 2025 image model in Versely
The all-in-one AI studio for creators. 60+ models for video, image, voice, music and lipsync, in your browser or on your phone.
One account for the web and iPhone.