Search “Video”

Hao AI Lab announces that FastH3 Omni can run locally on DGX Spark, RTX GPUs, and Apple MLX

Hao AI Lab announced that FastVideo FastH3 Omni can run locally on NVIDIA DGX Spark, RTX GPUs, and Apple MLX. The lab's reported performance figures include: generating a 5-second 480p video from a single photo and a prompt takes about 3 minutes, and about 2 minutes on two Sparks; FastH3 T2AV takes 52.6 seconds on a single Spark and 30.7 seconds on two Sparks. The clips in the demo video were all generated by DGX Spark using an 8-step NVFP4 checkpoint. The lab also provided links to a blog, a Cookbook, and Hugging Face. There are currently no subsequent developments.

Rohan Paul evaluates the roughly 8-hour battery life figure for the Unitree H2-X

On October 22, 2026, Rohan Paul published an article evaluating the roughly 8-hour battery life figure claimed by Unitree for its H2-X humanoid robot. He noted that the figure measures standing time, and it remains unclear how long the robot can sustain actual tasks such as patrolling, combat, or carrying loads. The article also mentions that the standard H2 weighs 70 kg, has 31 joints, is equipped with a swappable 0.97 kWh battery pack, offers about 3 hours of battery life, and is priced at $29,900; in the launch video, the H2-X leaps down from a 4-meter platform, kicks through a scrapyard, and repels a club-wielding intruder. Rohan Paul believes early buyers are more likely to be venues and warehouses that need overnight staffing than households. At present, this evaluation appears only in his own article, and there are no other developments.

fal Launches HeyGen Video and HeyGen Voice

fal has announced that HeyGen Video and HeyGen Voice are now available on the fal platform. According to information released by fal, both models have built-in audio: HeyGen Video supports generating videos with dialogue, ambient sound, and sound effects from text, images, and reference videos, with clips lasting 5 to 15 seconds and up to 2K resolution; HeyGen Voice supports 39 languages, instant voice cloning, and offers a faster Turbo tier. Currently, this news comes from a single source, fal, and there are no subsequent developments or facts yet.

Nathan Lambert Completes Presentation on the State of Open Models

On October 22, 2026, Nathan Lambert said he had finally completed a comprehensive presentation on the state of open models. According to his description, the content covers the definition of open models, licenses, competitiveness, signs of usage, the impact of knowledge distillation, and how to balance the risks and benefits of openness. He also plans to record a video lecture soon and hopes to redo it several times a year. There are currently no subsequent factual developments.

GM launches self-developed AI assistant for OnStar

General Motors has released its self-developed OnStar AI Assistant, with plans to first equip it on the 2027 Cadillac Lyriq. According to a cnBeta report on October 22, 2026, General Motors also plans to launch a new version of the OnStar mobile app in 2027, replacing OnStar Guardian and the separate apps for each brand. The assistant is positioned to shift OnStar from an accident assistance call center to a connected intelligent platform that can proactively offer suggestions; vehicle cameras will also be integrated with OnStar, enabling real-time viewing of the scene outside the vehicle and automatic recording of video of abnormal events. The assistant is currently at the launch stage, and no further facts about its subsequent rollout have been disclosed.

Karl's AI Watts shares six ways to use Opus for animation

Karl's AI Watts released a video sharing six ways to use Opus for animation. Published on October 22, 2026, this nearly 7-minute video covers product promos, explainer animations, hand-drawn animation, 3D scenes, music videos, and techniques for using AI video to create smooth transitions. The author says you don't need to learn to code first, but you do need to clearly explain the effect you want; the video includes segmented chapters such as the code animation principles at 00:23. Currently this content comes from a single source, with no further developments yet.

Kling 4.0 launches on the fal platform

Kling 4.0 is now available on the fal platform. The company calls it a new-generation native multimodal video model, with improvements in complex motion stability, consistency of character and scene references, and precision of cinematic narrative control. The model is currently available only through fal UX and fal Agent, with API access coming soon.

Sundar Pichai says Argon usage reached nearly 25 trillion tokens within 72 hours of launch

Sundar Pichai said that within 72 hours of Google's Argon going live, usage approached 25 trillion tokens, roughly 3 times the usage of Gemini 3.0 over the same period. He also mentioned that the model is currently being served, but compute capacity is limited. The remarks came from a video on Alex Kantrowitz's YouTube channel.

Meshy introduces Make Printable to thicken thin walls and repair holes

On October 21, 2026, Meshy introduced its Make Printable feature. According to a post published by Meshy, the feature only thickens weak areas in the model while preserving the remaining details, and supports one-click repair of holes and non-manifold edges. The post also includes a complete tutorial video from a 3D printing expert. Currently, the feature is still at the introduction stage, and there are no subsequent development facts.

The Flatkey gateway supports calling multimodal models in Codex and Claude Code via CLI

As a unified API gateway, Flatkey supports calling multimodal models in Codex and Claude Code via CLI. According to a report dated October 21, 2026, the gateway can call text models such as GPT, Claude, DeepSeek, GLM, and Kimi, as well as image and video models such as GPT Image and Seedance; users can access 100+ official models and 1,000+ tool APIs with a single Key, with costs consolidated into one bill. The report also states that its everyday prices are at least 20% lower than official pricing; during the October 19–23 promotion, GPT, GLM 5.2, Kimi K3, and DeepSeek are 40% off, Claude is 20% off, and Seedance 2.0 is 15% off. There are currently no subsequent developments.

Krea promotes Krea 3 for product image generation

Krea officially announced on October 22, 2026 that Krea 3 can be used for product image generation, quickly producing various angles of a product at a fraction of the cost of conventional solutions. The official tweet includes a trial link and a demo video, but no specific pricing or speed data. There are currently no subsequent developments; the specific pricing and generation speed of this capability await further official disclosure.

Artificial Analysis reviews the second anniversary of Video Arena's launch

Artificial Analysis published an article on October 22, 2026, reviewing the second anniversary of Video Arena's launch. The article states that the leaderboard initially covered only text-to-video and included 8 models; it later expanded to leaderboards for image-to-video, video with audio, and video editing, and has cumulatively evaluated more than 140 models. There are currently no further facts on subsequent developments.

User uses MiniMax M3.1 Flash to generate an Endurance browser artwork

On October 21, 2026, user Fengdu fengdu had MiniMax M3.1 Flash run continuously for 5 hours, building the Endurance spacecraft from Interstellar into a browser. According to a report by A Yi AI Notes, the work consists of 12 main cabin segments arranged in a ring, with a warm white outer shell and a dark gray skeleton, set against a blue planet background, and supports real-time rotation and camera movement. The report also states that the model supports million-token context and image and video understanding, with strong coding and multi-step execution capabilities; under official API pricing, each million output tokens cost $0.6, and cached reads cost $0.015. There are currently no subsequent development facts.

Kling AI announces full availability of KLING 4.0

On October 15, 2026, Kling AI announced the full availability of KLING 4.0, which users can access through a redesigned creation interface and which supports integration into video creation workflows via MCP and CLI; the company also previewed upcoming 4K and 1080p 10-bit HDR video output, as well as API access for enterprise customers and developers. On October 17, fal previewed the imminent launch of Kling 4.0, saying it can generate videos up to 30 seconds long in a single pass, supports native audio, and can use up to 15 reference materials to guide characters, scenes, and camera movement. On October 20, according to The Decoder, Kling AI released version 4.0, which can generate videos up to 30 seconds long in a single pass, supports input of up to 10 keyframe images, extends prompt understanding to 8,000 tokens, adds dual-channel stereo audio and lip sync in more than ten languages, and introduces Kling 4.0 Flash, a lightweight version supporting up to 720p and 20 seconds. On October 22, Kling announced the official opening of the KLING 4.0 API, allowing developers to integrate it into their own products and workflows, but did not disclose details such as interface capabilities, pricing, or availability scope.

Floo launches one-stop AI creation service platform

On October 21, 2026, Floo launched a one-stop AI creation service platform. According to a same-day report by China.com Technology, the platform links scriptwriting, character design, content production, and distribution and monetization into a complete chain, rather than only providing generation tools. The report states that the platform integrates nearly 10 SOTA language models, with the full version of Kling 4.0 debuting first across the internet at as low as ¥0.18 per second, and enables one-click distribution to Hongguo Short Drama, Douyin, and WeChat Channels, with a limited-time 100% revenue share to authors in Q4. There are currently no subsequent development facts.

NVIDIA announces global telecom operators are turning networks into distributed intelligent platforms

On October 21, 2026, NVIDIA announced at GTC Berlin that multiple telecom operators worldwide are extending NVIDIA accelerated computing to central offices, metro sites and base stations, turning networks into distributed intelligent platforms. Reports the same day show that T-Mobile, together with NVIDIA and Nokia, completed an AI-RAN call on a commercial network in Bellevue, Washington; BT, Deutsche Telekom, Elisa and Orange are conducting live network trials based on Nokia's CUDA-accelerated anyRAN software and NVIDIA Aerial RAN Computer. NVIDIA also launched two configurations, ARC MGX 2U and MGX 4U, for high-capacity sites, and Nokia and Cisco have integrated the Nemotron open model into their respective network automation software. Other reports say nearly 20 operators have announced AI-RAN collaborations based on Nokia anyRAN software and NVIDIA Aerial RAN Computer; BT and Deutsche Telekom plan to conduct live network trials in 2027, and Elisa is already running voice, mobile data and video analytics simultaneously in a live 5G spectrum environment.

Voxel51 releases version 3.0 data platform

On October 21, 2026, Voxel51 released version 3.0 of its data platform, targeting robotics and physical AI scenarios. According to Robotics & Automation News, the platform integrates the search, organization, annotation, and evaluation of robot multi-camera, video, and sensor data into a single platform, and adds a new AI assistant, Voxel51 Agent, that can execute tasks across the development workflow. The platform supports semantic search, text queries, and metadata filtering, as well as synchronized viewing of multimodal time series, and provides 2D, 3D, and video annotation and Agentic Labeling automatic annotation. The report also states that Voxel51 has partnered with Nebius to run its workflows on Nebius AI Cloud, and offers a free self-service version. There are currently no subsequent development facts.

Meta is developing Confidential VMs functionality on the Muse web client

Meta is developing Confidential VMs functionality on the Muse web client, allowing users to run Muse tasks in an encrypted confidential environment. Meta previously previewed that this feature would launch before the end of 2026. A newly surfaced hidden setting shows that the feature will initially be available only to Meta AI subscribers; after toggling it on, voice mode, image and video generation, and wearable device support will be temporarily unavailable. Meta says conversations run in a secure encrypted environment and that Muse data will not be used to improve Meta's AI. The feature is currently still in development, and there is no further evidence of an official launch date.

VAST open-sources real-time world model EdenLive

English text is being prepared.

English translation is being prepared.