Q2 2025 in AI video release notes
Q2 2025: 22 dated entries from 9 lines, 1 April 2025 to 30 June 2025. The table lists each; the chart counts them by kind of change. As of 2026-09-25.
A quarter of new models: Gen-4 at Runway on 1 April with a Turbo tier six days later, Veo 2 generally available in the Gemini API, Kling 2.0 Master in May and 2.1 in June, Hailuo 02 with 1080p and 10-second clips, Midjourney's first video model on 18 June, and Amazon's Nova Reel 1.1 in April with multi-shot videos of up to two minutes. Between them sit Gen-4 References, Luma's Reframe and Modify Video, Wan 2.1's first-and-last-frame and VACE models, and Veo 2's last frame on Vertex AI.
Inclusion rule. Kinds with at least one entry this quarter; one entry can count under several. Order. The order kinds take on every page.
1Every entry
| Date | Line | Version | What changed |
|---|---|---|---|
| 1 April 2025 | Runway | Gen-4 | Gen-4 is released for paid plans |
| 7 April 2025 | Amazon Nova Reel | Nova Reel 1.1 | Nova Reel 1.1 improves quality and latency for 6-second single-shot clips over Nova Reel 1.0 |
| 7 April 2025 | Amazon Nova Reel | Nova Reel 1.1 | Multi-shot videos up to two minutes, in an automated mode from one prompt or a manual mode with a prompt per shot |
| 7 April 2025 | Amazon Nova Reel | Nova Reel 1.1 | Manual mode, also called storyboard mode, takes a separate prompt for each shot |
| 7 April 2025 | Amazon Nova Reel | Nova Reel 1.1 | The user guide records video generation up to two minutes long |
| 7 April 2025 | Runway | Gen-4 | Gen-4 Turbo, the faster tier, opens to all users |
| 7 April 2025 | Wan | Wan 2.1 | Wan 2.1 start-and-end-frame model makes a transition between two images |
| 9 April 2025 | Google Veo | Veo 2 | Veo 2 becomes generally available in the Gemini API for text and image input |
| 18 April 2025 | Luma Ray | Ray2 | Nine Camera Angle Concepts are added for Ray2 |
| 29 April 2025 | PixVerse | Not named in the note | PixVerse MCP server released |
| 30 April 2025 | Runway | Gen-4 | Gen-4 References keeps characters and locations consistent |
| 9 May 2025 | Luma Ray | Not named in the note | Reframe outpaints a video or image beyond its borders |
| 13 May 2025 | Kling AI | Kling 2.0 | Kling 2.0 Master for text-to-video and image-to-video |
| 14 May 2025 | Wan | Wan 2.1 | Wan 2.1 VACE combines local editing, repainting, outpainting, extension and image references |
| 2 June 2025 | Luma Ray | Not named in the note | Modify Video keeps a performance and changes the world around it |
| 18 June 2025 | MiniMax Hailuo | Hailuo 02 | Hailuo 02 generates 1080p and clips of up to 10 seconds |
| 18 June 2025 | Midjourney video | Midjourney V1 video | The V1 video model animates images, with low or high motion and extensions of about 4 seconds |
| 19 June 2025 | Kling AI | Kling 2.1 | Kling 2.1 image-to-video in standard and pro modes, plus a 2.1 Master |
| 20 June 2025 | PixVerse | Not named in the note | Text-to-video and image-to-video take a camera movement field |
| 23 June 2025 | Google Veo | Veo 2 | Veo 2 on Vertex AI takes a last frame as well as a first, or a video to extend |
| 25 June 2025 | Runway | Gen-4 | An update to Gen-4 References improves object consistency |
| 30 June 2025 | Kling AI | Not named in the note | Lip-sync videos can run 60 seconds instead of 10 |
Inclusion rule. Entries whose note is dated in this quarter; undated version rows are not in any quarter. Order. By date, oldest first.
2Lines with entries
- Amazon Nova Reel: 4
- Google Veo: 2
- Kling AI: 3
- Luma Ray: 3
- Midjourney video: 1
- MiniMax Hailuo: 1
- PixVerse: 2
- Runway: 4
- Wan: 2