Search “Multimodal”

English translation is being prepared.

OpenAI Team Responds to Computer Use Progress Doubts, Recaps One-Week Sprint on Decisions API

In a podcast after OpenAI DevDay, Ari Weinstein, who is in charge of the Computer Use agent, responded to Dwarkesh's doubts about Computer Use progressing slowly, saying it is now completely different from a few months ago because the model has become better at debugging and retrying, and can write JavaScript to execute multiple steps at once while combining screenshots, accessibility, DOM, and Playwright. He also mentioned that Dots provides each Dot with a cloud Linux virtual computer, and that Computer Use has entered the Agents API. Nikunj Handa, head of the OpenAI API team, introduced updates to GPT-6 such as asynchronous tool calling, mid-turn steering, WebSockets, UltraFast inference, prompt caching and pre-warming, and compaction, and said the Decisions API is based on Luna weights, with no new model trained, and took only about a week from the emergence of the idea to a prototype.

English translation is being prepared.

After Meta's personal agent Muse went viral, who is most likely to build a Chinese version of Muse

Meta's personal AI agent Muse launched on September 8, surpassing 2.5 million downloads in 13 days, with 642,000 daily active users on the US mobile end, nearly 3 times that of ChatGPT in its early launch period; on September 21, Meta's market value rose by more than $190 billion in a single day. The article benchmarks domestic players across four items—entry point, payments, ecosystem, and execution capability—arguing that Tencent's Xiaowei has the most similar foundation but is constrained by WeChat's boundaries, Alibaba's Qwen is closest to transactions but lacks lifestyle scenarios, and ByteDance's Doubao phone relies on simulated operations and faces platform risk controls; in the short term, it is more likely that butlers within each ecosystem will coexist.

Building Real-Time Speech Applications on SageMaker AI with vLLM-Omni (Part 1)

AWS has published a tutorial demonstrating how to deploy Qwen3-TTS on Amazon SageMaker AI using the vLLM-Omni Deep Learning Container, enabling text streaming in and 24 kHz PCM audio chunks streaming out through SageMaker bidirectional streaming connections, along with a companion Gradio application. This DLC is based on vLLM-Omni v1.5 and adds the bidirectional streaming capability labels and routing middleware required by SageMaker AI; the example uses the native WebSocket route v1/audio/speech/stream.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

An agentic conversational video intelligence solution built on AWS

AWS has released an agentic video intelligence solution based on the Strands Agents SDK, in which a single AI agent orchestrates Amazon Bedrock, Amazon Rekognition, and Amazon Transcribe at runtime, allowing users to query video content by asking questions in natural language. Responses for already-analyzed content are returned within 1 second, while the first analysis of a new video takes 5–10 minutes; one media and entertainment company used it to reduce the manual review time for more than 200 multi-hour recordings by about 80%.

Ant Ling open-sources two 6B models in the Ming-Image-0.1-Design series

Ant Ling has open-sourced the Ming-Image-0.1-Design series, comprising two 6B models: Design generates complete visual designs such as UIs, infographics, and posters from text requirements, while Layer decomposes a design image into 2–9 semantically independent RGBA layers; Design Skill and PPT Skill are also open-sourced. In the UI/UX Design evaluation updated by Artificial Analysis on September 18, 2026, Design ranks 1st among open-source models with an Elo score of 1082; Layer achieves the best results on all 12 metrics of Crello-Test, reducing single-inference time from 795 seconds to 183 seconds. After Ant's internal Huamei integrated Layer, the time to first obtain editable layers dropped from over 5 minutes to about 40 seconds.

Evaluating Agents Equipped with Skills Using Strands Evals and Amazon Bedrock AgentCore

The Strands Evals SDK and Amazon Bedrock AgentCore Evaluations add three new skill-oriented evaluators: Skill Selection Accuracy returns a binary result for each invoked skill, Skill Instruction Following gives a five-level rating for each prescribed step, and Strands Evals also provides a deterministic Skill Invoked check. These evaluators can run on trajectories recorded by Strands Evals or on OpenTelemetry traces, and are used to pinpoint two types of failures: selecting the wrong skill and incompletely executing instructions; all are operated through the AgentCore CLI.

Debating RSI, the China-US gap, and uneven model capabilities with Epoch AI's JS Denain

In episode 19 of the Interconnects podcast, Nathan Lambert and JS Denain, head of the insights team at Epoch AI, discuss RSI, the China-US model gap, and uneven capabilities. JS believes the evidence in public blog posts from OpenAI and Anthropic is insufficient to prove that AI is about to self-accelerate, but the 2x monthly growth in Codex research spending mentioned in OpenAI's blog post is worth noting; both hold highly uncertain views on where AI is heading.

TypeSafe CEO Diogo Almeida on Jev: Building a System One Model for Production, Not for God

InstructGPT paper author and TypeSafe CEO Diogo Almeida detailed his new model Jev on the Latent Space podcast, describing it as a System One large programmatic model oriented toward code consumption, with its core innovation being an unpublished RLCD (reinforcement learning for calibrated decision-making), optimized for "intelligence per dollar" rather than human preference feedback. He criticized RLHF for causing mode collapse, sycophancy, and hallucinations, arguing that safety alignment suits products like ChatGPT but fragments intelligence at the API layer, and said TypeSafe would rather be seen as a data lab, refusing to publish benchmark tests. He revealed that Jev's daily token consumption has exceeded one trillion since launch, and previewed follow-up models such as ReasoningJev.

English translation is being prepared.

Google Research uses generative UI to let teachers create interactive learning simulations

Google Research has announced a new research experiment that lets teachers use generative UI (GenUI) to dynamically generate interactive educational simulations aligned with teaching objectives and curricula, and has released a library of more than 30 English-language interactive learning examples for middle and high school STEM, generated by AI and reviewed by teachers. The experiment is based on LearnLM, Google's education-specific model series released in 2024, and the generation process includes built-in pedagogical guardrails and a self-correction loop to check teaching objectives, solvability, and interface visual quality. Schools using Google Workspace for Education can sign up to provide feedback through the Google for Education Pilot Program.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.