AI daily

2026-07-23

OpenAI Launches Health Feature and Desktop Voice; Google Releases Gemini 3.6 Flash; GPT-5.6 Solves Math Conjecture

OpenAI expands ChatGPT with health data integration and desktop voice control, Google unveils three new Gemini models, and GPT-5.6 disproves a 30-year-old math conjecture with a 58-word prompt.

Back to feed
Selected
16
Sources
17
Featured
12
More
4
Previous2026-07-22Next
All reports

AI DAILY BRIEFING

OpenAI launched Health in ChatGPT, allowing eligible U.S. users to connect medical records and Apple Health data for personalized insights, and also rolled out ChatGPT Voice on desktop apps for voice-controlled multi-agent tasks. Google released Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber, improving agent efficiency and reducing API prices. In research, GPT-5.6 found a counterexample to the Dinitz-Garg-Goemans conjecture using only 58 words, and Tencent open-sourced three embodied foundation models with a 'three-layer brain' architecture.

OpenAI's health feature and desktop voice mark significant steps toward integrating AI into personal health management and everyday productivity.

Google's new Gemini models focus on practical improvements in speed, cost, and specialized domains like cybersecurity, intensifying competition in the AI model market.

GPT-5.6's mathematical discovery demonstrates LLMs' potential in scientific research, though the field remains cautious about reliability.

Watch next: Monitor adoption of OpenAI's health feature and desktop voice, Google's model performance benchmarks, and further AI-driven mathematical discoveries.

Featured

OpenAI Blog
6

ChatGPT Launches Health Feature

OpenAI has launched Health in ChatGPT, a new feature that lets eligible U.S. users securely connect their medical records and Apple Health data for personalized insights. This integration aims to help users better understand their health through AI-powered analysis.

SynthePulse InsightRead deep analysis
ChatGPT Will Give You Worse Health Advice If You Don't PayTHE DECODERChatGPT Health Feature Begins Rolling Out to U.S. UsersChatGPT (X)Health in ChatGPT Begins Rolling Out to U.S. UsersOpenAI (X)OpenAI Rolls Out ChatGPT Health to All US UsersThe Verge AIOpenAI makes ChatGPT Health available to all U.S. usersTechCrunch AI
少数派
3

Google Releases Gemini 3.6 Flash and Two Other Models

On July 21, Google released three new models: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber, focusing on improving AI agent efficiency, latency, and reliability. Gemini 3.6 Flash outperforms its predecessor on coding and computer operation benchmarks while reducing API prices. The Flash-Lite model targets high-throughput tasks, and the Cyber model is optimized for cybersecurity.

SynthePulse InsightRead deep analysis
Google Launches Gemini 3.6 Flash Model with Modest Improvements歸藏(guizang.ai) (X)Google Launches Three Gemini Flash Models, Focusing on Speed and Low CostAI前线
xAI Blog

Grok Build Launches Workflows Feature

Grok Build has introduced a new feature enabling the creation and execution of workflows, which are orchestration scripts that distribute tasks across hundreds of parallel agents, verify results, and provide consolidated reports in a single background run. This enhancement aims to improve efficiency and scalability for complex task automation.

SynthePulse InsightRead deep analysis
OpenAI (X)
2

ChatGPT Voice arrives in desktop app

OpenAI has launched ChatGPT Voice in the desktop app, enabling users to control their computer and direct multiple agents using voice commands. Powered by GPT-Live, the feature rolls out globally today on macOS and Windows to Plus, Pro, Business, Edu, and Enterprise plans.

SynthePulse InsightRead deep analysis
OpenAI launches voice control for Codex and ChatGPT desktop appsOpenAI Developers (X)
歸藏(guizang.ai) (X)
2

Black Forest Labs Releases and Open-Sources Flux 3 Video Generation Model

Black Forest Labs has released the Flux 3 video generation model and plans to open-source it. The model supports text-to-video, image-to-video, video reference generation, audio extension, keyframe control, multilingual dialogue, smart editing, text animation, and image output. Early access is available by application, with API and open-source release coming in the next few weeks.

SynthePulse InsightRead deep analysis
Flux 3 generates videos with native audio up to 20 seconds long, a first for Black Forest LabsTHE DECODER
机器之心

ECCV 2026 | AirZoo: Turning Global 3D Maps into Drone Training Grounds, a Unified Benchmark for Aerial Geometric 3D Vision

National University of Defense Technology and partners have released AirZoo, a large-scale unified dataset and benchmark for aerial geometric 3D vision. Covering 22 countries and 377 weather-condition trajectories, it includes over 1.2 million frames of pixel-level RGB-D data with 6-DoF geolocation annotations, supporting aerial image retrieval, cross-view matching, and multi-view 3D reconstruction. Experiments show that fine-tuning mainstream models on AirZoo significantly improves performance on aerial geometric 3D vision tasks. The work has been accepted by ECCV 2026.

SynthePulse InsightRead deep analysis
量子位

GPT-5.6 Overturns 30-Year-Old Math Conjecture with Just 58 Words of Prompts

Researcher Dmitry Rybin used GPT-5.6 Pro to find a counterexample to the Dinitz-Garg-Goemans conjecture, a longstanding problem in graph theory, using only 58 words of prompts. The AI constructed a small graph with 7 nodes and 9 edges that disproves the conjecture's claim about simultaneous cost and load optimization. This achievement highlights the potential of large language models in mathematical discovery and sparks discussion about AI's role in scientific research.

SynthePulse InsightRead deep analysis
AI前线

Tencent Open-Sources Three Embodied Foundation Models; Chief Scientist Zhang Zhengyou Explains How 'Three-Layer Brain' Solves Robot Latency

At WAIC 2026, Tencent open-sourced three embodied foundation models and introduced a 'three-layer brain' architecture that separates cognition, perception, and motor control to run at different frequencies, addressing robot response delays. Chief Scientist Zhang Zhengyou emphasized that the physical world operates on multiple time scales, making this layered design a long-term stable solution.

SynthePulse InsightRead deep analysis
Hacker News (AI filter)

Local AI that finds sensitive files on your Mac before attackers do

Vaultsort's Guardian is a local AI tool that scans Macs for sensitive files to prevent data breaches. It runs entirely on-device, ensuring privacy while identifying exposed documents before attackers can exploit them.

NVIDIA AI Blog

NVIDIA AI Supercomputer Comes Online at Naval Postgraduate School

NVIDIA CEO Jensen Huang commissioned a DGX GB300 supercomputer at the Naval Postgraduate School in Monterey, California, providing one of the world's most powerful AI platforms for military education and research. The system will support applications in weather prediction, cybersecurity, and disaster resilience, marking a milestone in the collaboration between NVIDIA and NPS.

SynthePulse InsightRead deep analysis
The AI Insider

Hadrius Raises $27M to Build Agentic Compliance Infrastructure

Hadrius, an agentic compliance infrastructure provider for financial services firms, raised $27 million in seed and Series A funding led by CRV, with participation from Y Combinator, Pathlight Ventures, and founders of Altruist, Jump AI, and FINNY. Serving over 500 financial institutions, the company reduces false positives by 95% and manual compliance work by 70%. The funding will accelerate product development and hiring to expand agentic oversight across compliance areas by 2026.

SynthePulse InsightRead deep analysis
The AI Insider

Multiverse Computing Reports All CompactifAI Models Now Run on Intel Xeon 6 Processors

Multiverse Computing announced that its CompactifAI-compressed Llama 3.3 70B model now runs on Intel Xeon 6 processors, achieving significant performance improvements. Benchmarks show up to 94% higher throughput and 48% lower latency compared to the uncompressed baseline, marking a step toward more energy-efficient AI deployment.

SynthePulse InsightRead deep analysis

More

机器之心

Kingsoft Office Launches Standalone AI-Native Office Agent 'Lingxi Pro'

At the 2026 AI Productivity Conference, Kingsoft Office unveiled 'Lingxi Pro', an independent AI-native office agent that reimagines office software by building AI from the ground up rather than adding AI to existing tools. It features project-based workflow, full deliverable generation (editable documents and PPTs), collaboration modes, and a unique 'auto-dreaming' memory mechanism to prevent context loss in long tasks, aiming to significantly boost productivity.

SynthePulse InsightRead deep analysis
The AI Insider

TYLsemi Announces $43M to Launch First Full-Stack Chiplet Platform for Custom AI Silicon

TYLsemi has emerged from stealth with $43 million in funding to develop the first full-stack chiplet platform for custom AI silicon. The company offers a portfolio covering IO, power delivery, and memory, with initial products sampling in 2027 via TSMC. Co-founded by semiconductor veterans, TYLsemi aims to reduce risk and time-to-market for AI infrastructure customers.

SynthePulse InsightRead deep analysis
The AI Insider

Feathery Raises $30M to Scale AI Operating & Decisioning System for Financial Services

Feathery, an AI operating and decisioning system for financial services, raised $30 million in total funding including a Series A round. The platform serves over 300 firms in insurance and wealth management, processing tens of millions of monthly submissions. The company will use the funds to expand products leveraging its cross-client data network and hire across engineering and go-to-market teams.

SynthePulse InsightRead deep analysis