Search “Video”

Hao AI Lab supports MiniMax H3 Flex in FastVideo

On November 4, 2026, UCSD's Hao AI Lab announced that it would integrate Pika Labs' open-source MiniMax H3 Flex into FastVideo. According to its published information, the official open-source inference code takes 59.7 seconds to generate a 5-second 480p video with audio on a single GB200, while FastVideo BF16 takes 26.6 seconds, NVFP4 QAD takes 16.8 seconds, and FP4 attention takes 15.2 seconds. Hao AI Lab claims it is up to 7 times faster than the official code, and added multi-GPU support, with 4 GPUs generating a 15-second clip 7 times faster. NVFP4 QAD distills BF16 Flex, coming 14-17% closer to BF16 than ordinary quantization; FP4 attention reduces a 768p 15-second clip from 101.8 seconds to 85.3 seconds. That clip takes 59.5 seconds on a single DGX Spark. FastVideo also supports the Open-H3-Regenerate-2K 2K model, and the weights, inference code, and complete training recipe have all been released.

Hao AI Lab announces that FastVideo supports Open-H3-Regenerate-2K

On November 4, 2026, UCSD Hao AI Lab announced that FastVideo now supports the Open-H3-Regenerate-2K model from HaoZhe65347 and NVIDIA AI, and provided a link to the 2K model. The same source also reviewed the series' earlier work: FastVideo optimized inference for Pika Labs' open-source MiniMax H3 Flex, generating a 5-second 480p video with audio on a single GB200 in 5.8 seconds, up to 7 times faster than the official open-source inference code, and added multi-GPU support. There are currently no subsequent developments.

Apple expands language support for the redesigned Health app

On November 3, 2026, Apple announced that the redesigned Health app expands language support to cover global English, Simplified and Traditional Chinese, French, Japanese, Korean, Norwegian, Portuguese, Spanish, Turkish, and Vietnamese, with Apple Intelligence at its core. The new version adds an Insights tab that summarizes data such as heart, sleep, and fitness metrics; the Longevity tab uses Health Age combined with Apple Watch metrics such as VO2 max and resting heart rate to reflect the body's condition. The update also introduces at-home fitness assessments, using the iPhone camera and Apple Watch to measure flexibility, strength, balance, and VO2 max on-device, without recording or storing video. According to Apple Newsroom, this version entered the iOS 27.3 developer beta that day and will be released for free later this year; 9to5Mac says the app was previously tested in the iOS 27.2 beta and will roll out to all users next month.

Karpathy builds HackerTube with Claude Opus 5.5 in two hours

Over the weekend, Andrej Karpathy used Claude Code together with Opus 5.5 to build the video site HackerTube in about two hours. According to a report by A Yi AI Notes, he filtered roughly 43 million Hacker News comments from the past twenty years down to 6,454 frequently recommended long videos; the site requires no login, uses a middle frame of each video as its thumbnail, and stores browsing state in the local browser session. He himself only manually applied for a Google Cloud YouTube API key, rented a VPS and bought a domain, then handed the IP over to Claude, which completed the deployment, security hardening and code optimization. A follow-up report by 36Kr · AI added that he used the desktop version of Claude Code with Opus 5.5 at Medium reasoning effort, that the page layout replicates YouTube, and that the underlying data can be exported and handed to one's own AI agent to rebuild a version. There is currently no further public progress on this matter.

Krea plugin supports turning sketches into cartoon animations within Grok

Krea announced that, with the Krea plugin, users can turn sketches into cartoon animations directly in Grok, simply by making a request to Grok. Previously, Krea had launched a plugin for the Grok bot that can generate images and videos within Grok. Currently, this feature has been integrated into Grok via the plugin, and users can use the sketch-to-cartoon-animation capability by making a request to Grok.

Nathan Benaich predicts AGI will be achieved by 2027

Speaking with Matt Turck on the MAD Podcast about the State of AI Report 2026, Nathan Benaich of Air Street Press reiterated his prediction that AGI will be achieved by 2027. He bases this judgment on the fact that AI can already approach his own level in knowledge work such as investing, AI research, finance, and law. He also discussed how images, video, and audio are converging into world models, as well as the cybersecurity risks posed by agents; ElevenLabs' agents already handle more than 15 million conversations per week, three times the volume in February. This prediction remains Benaich's personal judgment, and there are no subsequent developments of fact yet.

WorkBuddy integrates with Tencent Miora

As an AI workbench, Tencent WorkBuddy has been integrated with Tencent Miora, maintaining a consistent style across multimodal visual assets such as graphics and video, and supporting multi-window task execution. 36Kr·AI reported on October 30, 2026 that WorkBuddy can help vendors in the advertising and content industries break scattered briefs into communication plans, to-do lists, and PPTs, and organize guest schedules and budget tables; it comes with Skillhub, which has been integrated with operational skills for Xiaohongshu, WeChat Official Accounts, Douyin, and WeChat Channels, and can generate content in different platform sizes, styles, and copy. There is currently no follow-up development information.

Doubao launches daily check-in quota rewards and free trial for new users

On October 30, 2026, Doubao launched a daily check-in quota reward campaign for all users: each daily check-in earns 200 quota, doubled on Mondays, and 30 cumulative check-ins earn 8,800 quota, exceeding the 8,400 quota of the standard monthly plan. Reports on the same day also mentioned that new users logging into Doubao Desktop or Doubao Work Desktop for the first time can claim a 30-day free trial plan. The gifted quota is valid for 30 days and can be used for all capabilities except video generation and Pro models; check-in quota can be used immediately and is consumed first. Leiphone said this is the third time Doubao Work has distributed quota benefits to all users since its launch. There are currently no further developments.

Meta launches AI outfit app Style Sketch

Meta launches AI outfit app Style Sketch. According to a TechCrunch AI report, the app can connect to Instagram and build a virtual wardrobe from photos the user has already posted; it also supports manually adding items by taking photos, recording videos, or uploading album images. The app recommends outfits based on the user's style and lets users try them on a "realistic model" that matches their body type and skin tone; it also supports generating day-by-day outfit plans based on destination, trip length, and weather. Phoenix Technology reported the same day that the app has launched in Indonesia, Singapore, and Thailand, and supports iOS and Android. Meta did not respond on whether it will expand the release scope.

Rohan Paul said that Sunra resells multimodal model calls at the manufacturer's original price

Rohan Paul stated on October 29, 2026 that Sunra has released the Sunra API, through which users can call all major image, video, audio, and 3D models with a single key, billed at each model vendor's own API prices on a pay-as-you-go basis, with no subscription fee and no markup. Rohan Paul also said that Seedance 2.5 is priced on Sunra at half of what two other gateways charge, and that this unified entry point also makes it easier to run A/B tests on models from Google, OpenAI, ByteDance, and Kling. At present, this claim appears only in Rohan Paul's account, and there is no independent evidence from Sunra officially or from other sources.

Rohan Paul says Sunra resells multimodal model calls at the vendors' original prices

Rohan Paul stated on October 29, 2026 that Sunra has released the Sunra API, through which users can call all major image, video, audio, and 3D models with a single key, billed at each model vendor's own API prices on a pay-as-you-go basis, with no subscription fee and no markup. Rohan Paul also said that Seedance 2.5 is priced on Sunra at half of what two other gateways charge, and that this unified entry point also makes it easier to run A/B tests on models from Google, OpenAI, ByteDance, and Kling. At present, this claim appears only in Rohan Paul's account, and there is no independent evidence from Sunra officially or from other sources.

View story:Rohan Paul said that Sunra resells multimodal model calls at the manufacturer's original price

Articul8 launches decision model Bivia

Articul8 has launched Bivia, a multimodal decision model that can connect to AI agents, choose not to proceed when confidence is low, and weigh each step's decision against the overall task objective. The model supports structured responses, can be invoked mid-execution by an agent, and was trained on real production workflows. According to Testing Catalog, its input price is $0.042 per million tokens, with no subscription fee. Another report says Bivia can process documents, images, audio, video and time-series data, make continuous decisions, and defer to human judgment when uncertain; Articul8 grew out of a project delivered by an Intel team to enterprise customers one month before ChatGPT's release, is valued at $500 million, and says it has accumulated more than three years of production deployment experience in industries including semiconductors, energy and manufacturing.

Grok Bot updates this week: Pages, Vault, and 1Password integration

Grok Bot updated several features this week. According to a report by Ayi AI Notes, the update includes Pages, which can directly generate editable report pages and dashboards; Vault, used to store login and bank card information; and integration with 1Password. Vault and 1Password are designed to let the Bot autofill login and checkout information, but the Bot itself cannot see passwords or card numbers. The update also adds mobile app authorization, covering calendar, contacts, location, and alarms; on iPhone, it can manage reminders and control Home lights. In addition, there is video generation from descriptions or images, new Bot characters, and X Spaces. Currently, this report is the only source, and there have been no further developments.

Google Releases EmbeddingGemma 2 Paper

Google has released the EmbeddingGemma 2 paper, introducing an Apache 2.0-licensed on-device multimodal embedding model that maps text, code, images, audio, and video into the same vector space. Built on Gemma 4, the model learns from the larger Gemini Embedding 2 and is tuned with contrastive training; with 744M parameters, it outperforms models 2 to 12 times larger on multiple tasks. On the Pixel 11 Pro, the quantized text weights occupy about 191MB of active memory, while the full multimodal model takes about 567MB; the context window is expanded 4 times to 8K tokens, accommodating about 5.5 minutes of audio, 29 images, or 58 video frames. Embedding speech directly scores 69.54 on the MSEB spoken query retrieval benchmark, higher than the 57.59 achieved by transcribing first and then retrieving, and code retrieval on MTEB Code improves from 68.76 in EmbeddingGemma 1 to 78.68. There are currently no further developments.

Sam Altman says the world may need to limit the intelligence ceiling of a single AI model

Sam Altman proposed that the world may need to set a cap on the intelligence level of a single AI model. He stated that, at least for a period of time, and possibly for a very long time, people will have to reach some kind of limit on how smart a single model can be. He believes that a slowdown in progress would stem from this decision, rather than from AI lacking the ability to do research well. The relevant remarks come from the full video on the YouTube channel "Cheeky Pint." There are currently no follow-up developments.

meng shao says Video Agent products may be reduced to token relay stations

On October 29, 2026, Shao Meng put forward a judgment: although domestic Video Agent teams such as LibTV, TapNow, Flova, Pexo, and Medeo emphasize barriers such as ecosystem, Agent architecture, and industry know-how in financing roadshows, what they end up building may all turn out to be token relay stations. He speculated that if these websites supported BYOK, the cost for users to bring their own video generation model API Key would be lower than consuming Credits on the site, and even lower when the API Key has a discount. He suggested using REA or building one's own Video Agent from scratch paired with discounted API Keys to reduce costs, believing that for users making video content on TikTok, Ins, and Douyin, the money saved would be considerable. Currently, this judgment has no subsequent developments.

Kling AI announces that Kling4 is integrated into Genspark AI Media Studio

On October 29, 2026, Kling AI announced that Kling4 is now available in Genspark AI Media Studio. According to a message released by Kling AI, Genspark announced that Kling 4.0 has been integrated into the workbench, supporting the generation of 30-second videos in a single pass without stitching, and supporting up to 15 image, video, character, or voice references as well as 10 keyframes; credits are 50% off until November 4, and the official Kling AI account reposted this news. On the same day, QbitAI reported that Genspark officially launched the video Agent workbench Media Studio, integrating more than 70 model versions from over 20 vendors, covering video, image, audio, and 3D. This integration has now been publicly announced with the release of Genspark Media Studio.

Higgsfield AI launches Higgsfield Playable

Higgsfield AI announces the launch of Higgsfield Playable. Users can upload game videos, screenshots, or prompts to build mobile games with viral mechanics, and turn them into playable ads in the Enterprise plan. Currently, this news was released by Higgsfield AI, with no further developments yet.

VEED Dubbing launches on the fal platform

On October 28, 2026, fal announced that VEED Dubbing is now available on the fal platform. According to information released by fal, this feature can translate videos using cloned voices and lip sync; the Pro version supports 94 languages and 14 regional accents, preserves background audio and supports multiple speakers, and also provides controls for target accent and voice clone strength. There are currently no subsequent development facts.

Grok Bot updates this week: Pages, Vault, and 1Password integration

Grok Bot updated several features this week. According to a report by Ayi AI Notes, the update includes Pages, which can directly generate editable report pages and dashboards; Vault, used to store login and bank card information; and integration with 1Password. Vault and 1Password are designed to let the Bot autofill login and checkout information, but the Bot itself cannot see passwords or card numbers. The update also adds mobile app authorization, covering calendar, contacts, location, and alarms; on iPhone, it can manage reminders and control Home lights. In addition, there is video generation from descriptions or images, new Bot characters, and X Spaces. Currently, this report is the only source, and there have been no further developments.

View story:Grok Bot updates this week: Pages, Vault, and 1Password integration