Model releases and updates
The DecoderReported by 4 other sourcesMeta Launches Muse Ember, a Small Model for Routine Agent Tasks
Meta has introduced Muse Ember, its smallest and most efficient AI model to date, for routine agent tasks such as tool calls, small coding tasks, request routing, data extraction, and long document processing. It handles text, images, audio, and video, with a context window of up to one million tokens. Meta says it is available through the Meta Model API and Muse Code, priced at $0.10 per million input tokens and $0.50 per million output tokens.
ByteDance Releases Seedream 5.5 for More Stable Image Editing
ByteDance’s Seed team released the Seedream 5.5 image model on November 9, with editing that can keep unchanged areas pixel-level stable and generate images with transparent backgrounds. The post reports a severe text error rate of 0.4%, down from 2.8% in the previous generation, and says editing realism trails Nano Banana 2.1 by about 6%. Seedream 5.5 is rolling out across Jimeng, Doubao, and Doubao Work; developers can try Doubao-Seedream-5.5 on Volcano Engine Ark.
QbitAI Tests Tencent Hunyuan Hy4 on Multimodal Agent Tasks
QbitAI author Wen Le tests Tencent Hunyuan Hy4 through WorkBuddy on tasks including recreating a music player from a screenshot, building a 3D amusement park, creating and revising a 16-slide presentation, and making an image batch-processing tool. The article describes Hy4 checking its work, fixing issues, and rerunning tests during these tasks. It also reports internal evaluation results and says Hy4 is available through WorkBuddy and CodeBuddy for a limited free trial through November 15 at 23:59.
Product releases and updates
NVIDIA Introduces GR00T 2 and Robotics Technologies at CoRL
At CoRL, NVIDIA is introducing GR00T 2, an open world action model, alongside robotics simulation tools, benchmarks, and agentic workflows. The article reports that GR00T 2 achieved 71% average success across 50 tasks with one demonstration per task, rising to 95% with five; Isaac Lab 3.0 with Newton reached up to 2.5x the training throughput of Isaac Lab 2.3 with PhysX in a vision-based dexterous manipulation benchmark.
Qianwen Office 2.0 Launches Team Agent Collaboration Features
Qianwen Office released version 2.0 on November 9, adding a collaboration feature for people and Agents to work together in shared spaces. The update also includes automatic tasks triggered by events, in-chat document editing, a searchable materials space, and task and conversation sync across mobile, desktop, and web. A retail operations team uses the shared workspace to collect information across store group chats and update an AI spreadsheet.
Sundar Pichai · XSundar Pichai Says Google’s TPU Co-Design Speeds Hardware Innovation
Sundar Pichai says Google can innovate quickly for customers by having hardware engineers co-design systems alongside Google DeepMind. He contrasts this approach with waiting years for a traditional semiconductor cycle and notes that four distinct TPU systems were introduced within one generation in one year, attributing that milestone to a quoted post by Amin Vahdat.
Runway · XRunway Introduces Sightline for 360-Degree Panoramas and Shot Framing
Runway introduces Sightline, which creates a 360 panorama of any location and lets users frame shots from any angle while keeping the environment consistent across them. Users can export the panorama for 3D use as an environment in their scenes, and get started in App Studio.
Research
Elith and NVIDIA Release a Synthetic Finance Persona Dataset for Japan
Elith and NVIDIA released Nemotron-Personas-Finance-Japan, an open synthetic dataset designed for developing and evaluating AI for Japanese financial use cases. It contains 9 million personas across 50 fields, drawing on official statistics and household finance and financial literacy data. The dataset is available for commercial and non-commercial use under the CC BY 4.0 license.
Tips and perspectives
OpenAI Puts a $200 Codex Edition Chromatic Handheld on Sale
OpenAI has listed a Codex edition of the ModRetro Chromatic handheld for $200 on Supply Co. It includes a writable cartridge and a developer mode activation code; the accompanying Codex plugin can help create Game Boy games that can be played in a browser emulator or written to a cartridge. The edition ships to the United States, Canada, Australia, New Zealand, Japan, the United Kingdom, and the European Union.
Baidu Describes ERNIE Evolve’s Role in Developing ERNIE 5.5 Flash
Baidu has disclosed that its ERNIE Evolve automated research system is being used to optimize inference for the upcoming ERNIE 5.5 Flash. The article reports speed or throughput gains across three tasks, including generation speed reaching 623.6% of baseline in a Kunlunxin P800 low-latency setting; Baidu says the changes passed stress tests, correctness checks, and production validation. A source close to Baidu said the new model may be released soon.
AWS Machine Learning BlogFirst-party
Amazon Bedrock Managed Knowledge Base Adds Single-Shot and Retrieval-Only Modes
Amazon Bedrock Managed Knowledge Base now supports single-shot RAG and retrieval-only requests through the AgenticRetrieveStream API, alongside existing agentic retrieval. Setting maxAgentIteration to 1 retrieves once and generates an answer; setting it to 0 returns passages without a language model call. Values from 2 to 5 retain agentic retrieval, with up to n minus 1 planning rounds before answer generation.
Tomorrow at 08:00, a briefing like this arrives in your inbox.
The same kind of briefing as the one above: a few stories worth reading. Free. Mail starts after you confirm, and every letter has a one-click unsubscribe. Read a past issue · 2026-11-09