Flash

The news, the context, and what changes.

RSS
Flash

Volcengine Launches Premium Video Generation Model Seedance 2.5 4K

Volcengine has released the premium video generation model Seedance 2.5 4K, focusing on optimizing facial expressions, subtle movements, and visual details. The model supports 10-bit color depth, significantly enhancing texture quality, color layers, and gradient effects. It also maintains better clarity and consistency of materials during reference generation and video editing tasks. The workflow allows users to generate 480P drafts for storyboard verification before producing the final 4K vide

88Z AI editorial1 min read
Flash

MiniMax Open-Sources MiniMax M3.1 Flash Model Weights

MiniMax has open-sourced the weights for its MiniMax M3.1 Flash model. Featuring 428 billion total parameters and 23 billion activated parameters, it supports a context window of one million tokens. Within 72 hours of launch, it processed over 100 trillion tokens, with inference costs cut by more than half compared to the previous generation. It ranks first among domestic models in several Agent benchmarks. Ten chip manufacturers, including Huawei Ascend and NVIDIA, along with inference framewor

88Z AI editorial1 min read
Flash

iFlytek Launches Edge-Cloud Integrated Spark X2.6 and Related Products

iFlytek has launched the edge-cloud integrated Spark X2.6, SparkAudio 1.0 voice foundation model, and embodied multimodal interaction technology. The company also introduced AI glasses, Robot Super Brain Platform V3.0, and other software and hardware products. Emphasizing full-stack domestic computing power and edge-cloud synergy, iFlytek focuses on security, privacy, and cost-effectiveness, with applications in education, healthcare, law, and the developer ecosystem.

88Z AI editorial1 min read
Flash

Tencent and Fudan University Launch WorkForge Framework for Agent Training Environment Construction

Tencent Hunyuan and Fudan University jointly launched the WorkForge framework, designed to scale the construction of verifiable agent work environments from real-world resources. They simultaneously released the WorkForge-16K dataset, covering 16,700 work environments across 40 professional fields. The WorkForge-27B model, trained on real workspace trajectories from this framework, performed excellently, comprehensively surpassing models with hundreds of times more parameters in five major work

88Z AI editorial1 min read
Flash

OpenAI and Google Cloud Announce Partnership, GPT-6 Astra to Launch on Model Garden

OpenAI has partnered with Google Cloud to bring its advanced models and Codex agent workflows to Google Cloud Model Garden, with a launch expected within weeks. Enterprise customers can count OpenAI model costs toward their GCP committed spend while leveraging Google Cloud's data governance and high-performance infrastructure. OpenAI models will also be integrated into Google Antigravity and Gemini Agent in the future.

88Z AI editorial1 min read
Flash

Kling 4.0 Full Version Officially Launches on LibTV

The full version of Kling 4.0 has launched on LibTV, supporting 30-second native video generation, enhanced physical consistency and director-level control, along with upgraded audio-visual performance.

88Z AI editorial1 min read
Flash

Anthropic Launches AI Agent Claude Duet

Anthropic has launched Claude Duet, an AI agent for SMEs and professionals built on the Opus 5.5 model. It features daily briefings, app monitoring, dashboard creation, meeting scheduling, and financial forecasting, with an independent browser VM for background operation, directly competing with OpenAI Dots and Meta Muse.

88Z AI editorial1 min read
Flash

Memory Tensor Open-Sources Memory-Native Foundation Model MemT 1.0 Series

The Memory Tensor team has open-sourced the MemT 1.0 series of foundation models, including Flash and Nano versions. This series integrates updatable internal memory into computation, allowing access to historical information without retrieving original text. In 11 memory benchmarks, the two models achieved average scores of 50.6 and 47.0, ranking top two among parameterized memory methods. Technical highlights include Gated DeltaNet-2 per-token updates, MoS hybrid storage expansion, and a two-s

88Z AI editorial1 min read
Flash

AI Meeting Assistant Otter.ai Raises $320 Million

Otter.ai announced a $320 million non-dilutive financing round from General Catalyst's Customer Value Fund, aimed at driving business growth, market expansion, and product innovation. The company had previously raised a total of $50 million, with last year's annual recurring revenue exceeding $100 million and its user base growing from 25 million to 40 million.

88Z AI editorial1 min read
Flash

Xingchen Lab Open-Sources Intent Understanding Model Xing4.0-29B-Intent

Xingchen Lab has open-sourced the intent understanding model Xing4.0-29B-Intent, releasing 2.75 million training corpus entries, model weights, and inference images. The model uses a unified semantic parsing framework to support intent classification and parameter extraction, converting natural language into structured JSON. Evaluations show a comprehensive score of 58, surpassing the top score of 49 achieved by 18 mainstream general large models.

88Z AI editorial1 min read
Flash

DeepSeek Open-Sources Production-Grade Inference Framework DeepInfer

DeepSeek has announced the full open-sourcing of its internal inference framework, DeepInfer. Integrating high-performance operators and an optimization engine, the framework supports distributed inference and advanced decoding techniques. It is compatible with NVIDIA and Ascend hardware and currently powers the high-throughput, low-latency services of DeepSeek V4.1 Flash.

88Z AI editorial1 min read
Flash

Free removal of TV ads and bloatware, open-source project CleanYourTV goes viral

CleanYourTV is an open-source Android TV cleanup tool that requires no root access. It connects via ADB to automatically disable ads, telemetry, and bloatware using AI assistants, replacing them with a lightweight desktop. With extensive device support and one-click restore, it has recently gained significant popularity.

88Z AI editorial1 min read