cho.sh

researchmini

6 alerts

My alerts

All alerts

AI video model releasesby User 6432 AI video model releasesby User 6432 Duneby 제욱 AIby 제욱 App Store 5.2.3by 주현 해커뉴스by 성현

Sign in to create alerts.

Lightricks Releases Apache 2.0 LTX-2.3 Open-Weight 4K Video Model · cho.sh

Lightricks Releases Apache 2.0 LTX-2.3 Open-Weight 4K Video Model

AI video model releases

Lightricks Releases Apache 2.0 LTX-2.3 Open-Weight 4K Video Model

Lightricks released LTX-2.3 on June 18, 2026 under the Apache 2.0 license, with downloadable weights, reference Python code, a ComfyUI node, and LoRA fine-tuning code.
The 13 billion-parameter diffusion transformer generates 4K clips up to 5 seconds long at 24 frames per second and supports both text-to-video and image-to-video workflows.
The article says LTX-2.3 needs at least 16GB VRAM for 1080p generation and 24GB VRAM for 4K, allowing local inference on a single consumer GPU.
Lightricks reports an 84.2 VBench-HD score and says the model was trained on licensed and public-domain footage; it also reports a 78.5 EvalCrafter text-alignment score.
The release has no hosted public demo, so users must download the safetensors weights and install the included tools locally.

4DAnyone generates 16 consistent views from one video for 4D human reconstruction Alibaba's Wan 3.0 beta claims 30-second single-pass AI video and document inputs Alibaba rolls out Wan 3.0 with 30-second 1080p document-to-video generation Alibaba releases Wan3.0 with 30-second generation and document-to-video input Alibaba Rolls Out Wan3.0 Video Model With 30-Second Generation and Document Inputs Alibaba launches Wan3.0 video model with 30-second generation and document inputs Lightricks Releases Apache 2.0 LTX-2.3 Open-Weight 4K Video Model Alibaba launches closed Wan3.0 video model with 30-second 1080p generation Alaya Lab releases Evoke, a 14B open world model with external scene memory for hour-long interaction Alibaba puts Wan3.0 video model into beta with 30-second multimodal generation Alibaba launches Wan3.0 with 30-second video generation from documents and slide decks LTX releases open-weight LTX-2.5 video world model with multishot generation and 6.8-second 720p clips Alibaba launches Wan3.0, which turns business documents into 30-second videos Alibaba launches Wan3.0 video model with 30-second outputs from documents and web pages MiniMax H3 and ByteDance Seedance 2.5 Launch Hours Apart Alibaba launches Wan3.0 video model with native 30-second generation and document inputs MiniMax and ByteDance Ship Rival H3 and Seedance 2.5 Video Model Updates Alibaba makes Wan3.0 generally available with 30-second multimodal video generation GIGAZINE title reports Alibaba's Wan3.0 video model release with 30-second generation Alibaba launches Wan3.0 beta for 30-second video generation from documents and mixed media Alibaba Cloud launches Wan3.0 video model with 30-second clips and document inputs Alibaba releases Wan3.0 video model with 30-second generation from documents and web pages Alibaba opens Wan3.0 beta with 30-second 1080p video and document-to-video inputs Luma releases Ray3.14 with native 1080p, claimed 4x faster generation and 3x lower per-second video costs Google releases Veo 3.1 and Veo 3.1 Fast in paid Gemini API preview Alibaba opens Wan3.0 beta with 30-second video generation and multimodal reference inputs ByteDance’s Seedance 2.5 claims 30-second AI video generation with native audio Lightricks Releases Open-Weights LTX-2.5 With Native 4K and Multi-Shot Video Generation Luma releases Ray3 video model with visual reasoning, 16-bit HDR EXR, and Adobe Firefly access WaveSpeedAI adds Vidu Q3 Start-End to Video with dual-keyframe control and 1080p output LTX-2.5 Open-Weights Video Model Targets Local NVIDIA RTX Inference IndiaAI-backed Avataar launches Varya video model, claiming four-step generation at ₹0.48 per second Alibaba Releases Open-Weight Wan 2.2 Video Models With Noise-Routed MoE Architecture ByteDance keeps Seedance 2.5 closed as MiniMax plans H3 weight release LTX releases open-weights LTX-2.5 video world model with robotics checkpoint Layer adds MiniMax H3 for stereo 2K text, image, and reference-to-video generation MiniMax Hailuo-2.3 launches in VEED with 1080p generation and a 55-second fast image model Alibaba releases Wan3.0 with 30-second generation and document-to-video input MiniMax launches H3 video model with open weights, undercutting rivals on price MiniMax launches open-weight H3 video model, undercutting rivals on 2K generation cost LTX-2.5 launches as open-weights video model with native ComfyUI integration and Nvidia GB200 speedups Lightricks Releases LTX-2.5, an Open-Weights Video World Model Accelerated by NVIDIA LTX-2.5 open-weights video model generates clips in 6.8 seconds, ships with ComfyUI integration Google updates Veo 3.1 Ingredients to Video with vertical output and 4K upscaling MiniMax launches H3 video model with 2K resolution, open weights coming ByteDance launches Seedance 2.0, an AI video model combining text, image, video, and audio prompts MiniMax releases open-weight H3 video model, undercuts ByteDance's Seedance 2.0 on price and some benchmarks Alibaba launches Wan3.0 AI video model in public beta with 30-second clips and multimodal input support MiniMax and ByteDance Both Ship AI Video Model Upgrades Within Hours of Each Other MiniMax launches H3 video model with 2K, 15-second clips and native stereo sound, plans open weights Kuaishou Launches Kling AI 3.0 Video Model With 15-Second Clips and Native Multilingual Audio xAI Releases Grok Imagine Video 1.5, Doubles Generation Speed With Fast Variant Alibaba launches Wan 3.0 in public beta, doubling clip length to 30 seconds ByteDance Launches Seedance 2.5 Video Model with 30-Second Clips, Skips US Market MiniMax H3 (Hailuo 3.0) Launches with Native 2K Video and Synced Stereo Audio in One Pass MiniMax Releases Open-Weight H3 Video Model, Blocks US, UK, EU and South Korea Over Copyright Suits MiniMax Launches Hailuo 2.3 Video Model with Improved Physics, Stylization, and Micro-Expressions Black Forest Labs launches FLUX 3 Video, claims SOTA in text-to-video and ties top model in image-to-video MiniMax releases 33B-parameter H3 video model weights, but keeps context module and 2K upscaler closed MiniMax H3 unifies text, image, video and audio into one AI model with native 2K output MiniMax releases open-weight H3 video model, day-zero support lands in ComfyUI for local use MiniMax releases omni-modal H3 video model with 2K, 15-second clips and native stereo audio, plans open weights ByteDance Launches Seedance 2.5 on Dreamina With 30-Second Clips and 50 Reference Assets MiniMax H3 becomes first open-weight AI video model to lead a benchmark ranking MiniMax open-sources H3, an omni-modal video model with native stereo audio up to 2K and 15 seconds MiniMax releases open-weights H3 video model with native stereo audio and 2K output, day-0 support in ComfyUI xAI's Imagine Video 1.5 Adds Image and Voice Reference Tools, Native 1080p Output MiniMax launches Hailuo S2V-01, a subject-reference video model for character consistency from one image MiniMax Launches H3 2K Multimodal Video Model, Open Weights Promised But Not Yet Released Alibaba's HappyHorse 1.1 video model hits No. 2 on Arena rankings as Sora and Seedance exit market ByteDance's Seedance 2.5 Doubles Video Length to 30 Seconds, Adds 50-Reference Input and Timestamp-Level Editing MiniMax Launches H3, an Omni-Modal Video Model Generating 15-Second 2K Clips With Native Stereo Audio MiniMax Launches Open-Weights Video Model H3 Same Day as ByteDance's Closed Seedance 2.5 ByteDance Launches Seedance 2.5 With 30-Second Native Clips and Maya, Blender Plugins MiniMax Launches Omni-Modal Video Model Generating 15-Second 2K Clips With Native Stereo Audio MiniMax Launches Hailuo 3.0 (H3), AI Video Model With Native Synced Audio and Multi-Reference Character Consistency MiniMax H3 Debuts 2K AI Video With Native Audio, Promises Open Weights xAI updates Grok Imagine Video 1.5 with image and voice references, text-to-video, native 1080p ByteDance launches Seedance 2.5 AI video model with multi-input generation MiniMax Launches H3, a Multimodal Video Model That Edits Footage and Generates Native Audio UCSD's Hao AI Lab releases FastWan, generating 5-second 480P video in 5 seconds via sparse distillation FastVideo releases FastWan2.1-T2V-1.3B, a 3-step distilled video model hitting 16 FPS on one H100 ByteDance Releases Seedance 2.5, First AI Video Model to Generate Native 30-Second Clips MiniMax and ByteDance Launch Rival AI Video Model Updates Hours Apart MiniMax launches H3 video model with 2K native-audio clips, undercuts rivals on price MiniMax releases open-weight H3 video model, undercuts ByteDance's Seedance 2.0 on price ByteDance Launches Seedance 2.5 With 30-Second Native Clips and 3-Minute Long Video Mode MiniMax Releases Multimodal Model H3, Generates 15-Second 2K Audio-Video at Under a Third of Rival Costs, Plans Open Source ByteDance releases Seedance 2.5, a video model generating 30-second clips in one run MiniMax launches H3 video model, plans open-weight release within days MiniMax launches H3 video model with open weights, undercutting rivals on price ByteDance releases Seedance 2.5 with 30-second one-pass video generation and multimodal referencing MiniMax and ByteDance Launch Rival AI Video Models Hours Apart ShengShu launches Vidu S1, a real-time interactive AI video model running on consumer GPUs Black Forest Labs launches FLUX 3, a single model generating video, audio and robot actions Kuaishou Launches Kling AI 3.0 with 15-Second Native-Audio Video and Multi-Shot Storyboarding NVIDIA's Cosmos-H-Dreams Distills Surgical World Model to 160 FPS on One GPU Black Forest Labs launches FLUX 3, one model that generates images, video, audio, and robot actions Alibaba Launches Wan2.2, First Open-Source MoE Video Generation Models With Cinematic Control OpenAI unveils Sora, a diffusion transformer that generates up to a minute of video from text