12 / 914 contents

GPT-Live 소개
News2026.07.13

GPT-Live 소개

1. 🎙️ OpenAI has launched GPT-Live, a next-generation voice model featuring a full-duplex architecture that enables natural, human-like interaction with real-time responsiveness. 2. 🧠 The system operates by utilizing a background frontier model, specifically GPT-5.5 at launch, to handle complex re

#LLM#Speech Model#AI Communication
igor
Introducing Muse Spark 1.1
News2026.07.11

Introducing Muse Spark 1.1

1. 🚀 Meta has introduced Muse Spark 1.1, an advanced multimodal reasoning model designed for complex agentic tasks, coding, and computer use with a 1-million-token context window. 2. 🛠️ The model excels at orchestrating multi-agent systems and navigating computer interfaces by automating scripts o

#Agent#AI Model#LLM
성산
GPT-5.6: Frontier intelligence that scales with your ambition
News2026.07.09

GPT-5.6: Frontier intelligence that scales with your ambition

1. 🚀 OpenAI has launched the GPT-5.6 model family, including flagship Sol, balanced Terra, and cost-efficient Luna, which offer significantly improved intelligence and efficiency across coding, research, and agentic workflows. 2. 🛠️ The new series introduces features like Programmatic Tool Calling

#Agent#AI#GPT-5.6
Homin.Lee
Introducing Muse Image and Muse Video
News2026.07.08

Introducing Muse Image and Muse Video

1. 🎨 Meta Superintelligence Labs has introduced Muse Image and Muse Video, advanced media generation models that utilize agentic tool use, such as search and coding, to enhance factual accuracy and image composition. 2. 🧠 Muse Image demonstrates emergent self-refinement capabilities through reinfo

#Agentic AI#AI#Content Seal
Homin.Lee
Orca — The most powerful Agent Development Environment (ADE)
Service2026.07.07

Orca — The most powerful Agent Development Environment (ADE)

1. ⚠️ The provided link does not contain a valid academic paper for analysis, as it leads to a broken download page. 2. 🔍 Please ensure you provide a direct link to a PDF or a copy of the text you wish to have summarized. 3. 📝 Once the correct document content is supplied, I will be able to genera

#ADE#Agent#Development Environment
Mineru
GitHub - Yeachan-Heo/gajae-code: Gajae Code MVP
Service2026.07.07

GitHub - Yeachan-Heo/gajae-code: Gajae Code MVP

1. 🦀 Gajae-Code is an experimental, external coding-agent harness designed to provide a structured workflow through deep interviewing, implementation planning, and evidence-based goal tracking. 2. 🛠️ The platform enhances developer productivity by supporting tmux-backed parallel execution, isolate

#Agent#AI#Automation
Mineru
Andrej Karpathy, Google and Garry Tan agree Markdown is the answer, but they're not solving the same problem
Blog2026.07.07

Andrej Karpathy, Google and Garry Tan agree Markdown is the answer, but they're not solving the same problem

1. 📝 Andrej Karpathy, Google, and Garry Tan have all independently identified Markdown as the preferred format for AI agent memory and organizational knowledge storage. 2. 🌐 While Karpathy focused on personal knowledge bases, Google introduced the Open Knowledge Format for enterprise context, and

#Agent#AI#Cloud Native
Homin.Lee
Meta's Watermelon AI model matches OpenAI's GPT-5.5 benchmarks
News2026.07.07

Meta's Watermelon AI model matches OpenAI's GPT-5.5 benchmarks

1. 🍉 Meta claims its experimental "Watermelon" AI model has reached performance parity with OpenAI’s GPT-5.5 based on internal, undisclosed benchmarks. 2. 💻 The model, which reportedly utilizes ten times the computational resources of its predecessor, remains in the training phase with no timeline

#AI#Benchmark#LLM
Homin.Lee
Meta is finally catching up to OpenAI, its AI leader says
News2026.07.07

Meta is finally catching up to OpenAI, its AI leader says

Unable to extract text from the document. It may be a scanned image PDF or a document without a text layer.

#AI#Meta#OpenAI
Homin.Lee
Cluedoc: document your codebase as interlinked visual papers
Paper2026.07.06

Cluedoc: document your codebase as interlinked visual papers

1. 🏗️ Cluedoc is an agentic tool that automatically generates and maintains human-readable, academic-style documentation for software features as a codebase evolves. 2. 🌳 It structures these documents into a hierarchical capability tree where each paper acts as a modular, visual explanation of a s

#Agent#AI#Codebase
Homin.Lee
Alibaba Unveils Agent Tool Selection Framework 'SkillWeaver'... "99.9% Token Usage Reduction" - AI Times
News2026.07.06

Alibaba Unveils Agent Tool Selection Framework 'SkillWeaver'... "99.9% Token Usage Reduction" - AI Times

Unable to extract text from the document. It may be a scanned image PDF or a document without a text layer.

#Agent#AI#Alibaba
Leon
Leanstral 1.5: Proof Abundance for All
Blog2026.07.06

Leanstral 1.5: Proof Abundance for All

1. 🚀 Leanstral 1.5 is a new, open-source 6B active parameter model that achieves state-of-the-art performance in formal verification, including saturating the miniF2F benchmark and solving 587 PutnamBench problems. 2. 🧠 The model utilizes a sophisticated three-stage training process involving mid-

#Code Verification#Formal Verification#Lean 4
Homin.Lee
Scroll to load more