Ghim khớp với tag:"Video Generation"

Google DeepMind releases Veo 3 at Google I/O 2025, the first Veo model to generate synchronized dialogue, sound effects and ambient noise with the picture, available to AI Ultra subscribers in the US through the Gemini app and the new Flow tool.

Kuaishou introduces the Kling 2.1 family in Standard (720p), Professional (1080p) and Master tiers, initially image-to-video only, days after Google's Veo 3; Master costs about $3 for ten seconds.

Google introduces Veo 3 Fast in the Gemini app and Flow, a variant more than twice as fast as Veo 3 at the same 720p resolution, with three Fast videos a day for AI Pro subscribers and 20 credits per clip in Flow.

ByteDance's Volcano Engine launches Seedance 1.0 Pro, a text-to-video and image-to-video model with native multi-shot storytelling at up to 1080p, at its FORCE conference alongside Doubao 1.6.

xAI rolls out Grok Imagine in the Grok iOS app to SuperGrok and Premium+ subscribers, turning generated or uploaded images into short clips with synced sound and four styles: Custom, Normal, Fun and Spicy.

Kuaishou launches Kling 2.5 Turbo, a faster and about 30% cheaper video model that Kling says beats Veo 3-fast and Seedance 1.0 in blind tests and that topped Artificial Analysis's text- and image-to-video arenas within ten days.

OpenAI launches Sora 2, its first video model with synchronized dialogue and sound, with an invite-only TikTok-style Sora iOS app where users make 10-second clips and insert themselves through 'cameos'.

At DevDay on 6 October 2025 OpenAI opens its Videos API to developers with sora-2 and sora-2-pro, per-second pricing from $0.10, and audio-synced video output.

xAI ships Imagine v0.9, a new video model with big gains over v0.1 in visual quality, motion and audio, including speech and songs that match lip movements, free on all of its products.

Google releases Veo 3.1 and Veo 3.1 Fast with richer audio, stronger prompt adherence and Flow upgrades: up to three reference images, first-and-last-frame transitions, scene extension and object insertion, in the Gemini app, Gemini API and Vertex AI.

Volcano Engine launches Doubao-Seedance-1.0-pro-fast, about three times faster than 1.0 Pro and 72% cheaper per 5-second 1080p clip.

Kling AI launches Video 2.6, its first model with native audio: voices, singing, sound effects and ambience generated in one pass with the picture, in English and Chinese, for 5 and 10 second clips.

ByteDance Seed releases Seedance 1.5 pro, its first native joint audio-video generation model, with multilingual and dialect lip-sync and cinematic camera control, available on Volcano Engine.

xAI opens the Grok Imagine API, a text-to-video, image-to-video and video-editing bundle with native audio that debuted first on Artificial Analysis's video leaderboards at $4.20 per minute.

xAI launches Grok Imagine 1.0, its 'biggest leap yet': 10-second videos at 720p with dramatically better audio, announced alongside a claim of 1.245 billion videos generated in the previous 30 days.

Kling AI launches the 3.0 series, including Video 3.0 and Video 3.0 Omni, with 15-second clips, native multilingual audio with accents and dialects, multi-shot storytelling and video-referenced characters, first to Ultra subscribers.

ByteDance releases Seedance 2.0 in China, a multimodal audio-video model taking text, image, audio and video references for 4 to 15 second clips; viral celebrity clips draw Disney and Paramount legal threats within days.

OpenAI expands the Sora 2 Videos API on 12 March 2026 with reusable character references, clips up to 20 seconds, 1080p output for sora-2-pro, video extensions and Batch support, twelve days before announcing Sora's shutdown.

Google launches Veo 3.1 Lite in the Gemini API and AI Studio, its cheapest video model at $0.05 a second for 720p with audio and the same speed as Veo 3.1 Fast, with 4, 6 or 8 second clips at 720p or 1080p.

xAI puts grok-imagine-video-1.5-preview in its API: an image-to-video model with native audio at up to 720p that debuted at or near the top of the Image-to-Video arenas, two days after a first appearance on Arena.

xAI takes Grok Imagine Video 1.5 out of preview in the Imagine API and rolls out a Fast tier in the apps that makes a 6-second 720p clip in about 25 seconds, down from 40-plus.

ByteDance releases Seedance 2.5, unveiled at Volcano Engine's FORCE conference on 23 June 2026: 30-second single-pass clips, up to 50 multimodal references and local editing, live on Jimeng and Doubao from 31 July.

xAI upgrades grok-imagine-video-1.5 with text-to-video, up to seven reference images, voice references and native 1080p, live in the xAI API and on grok.com/imagine.

Kuaishou's Kling 4.0 reaches full availability on 15 October 2026 after a 28 September Flash preview: native 30-second clips, 10 keyframes, 15 combined references, stereo audio with multilingual lip-sync and editing, with 4K and HDR still marked coming soon.

ByteDance has not announced a successor to Seedance 2.5, but its cadence and its twice-yearly Volcano Engine FORCE launches point to the next Seedance model around mid-December 2026; third-party 'Seedance 3.0' sites are unofficial.

xAI has announced no successor to Grok Imagine Video 1.5, but video releases have come every one to four months and its image side moved to 2.0 in August 2026, which points to a next Grok Imagine video model around mid-December 2026.

Kuaishou has announced nothing after Kling 4.0 (full release 15 October 2026), but its release gaps, a median of about three months, and pending 4K, HDR and API features point to the next Kling model around January 2027.

Google has not announced a Veo 4: a Polymarket market on its release (about $45,000 traded) resolved No at every deadline through 30 June 2026, and Google shipped Gemini Omni Flash at I/O 2026 instead; the next Google video model is expected at I/O in May 2027.