Aug 04, 2026 Technology Pulse

🧠 A unified API for AI model routing

Google Cloud API Gateway now offers a model routing feature in Public Preview, letting developers dynamically route traffic to models such as Gemini, Claude, or OpenAI OSS-GPT without hardcoding endpoints or managing open-source proxies. Routing rules can be configured in OpenAPI 3.x specifications by mapping virtual model names to specific backend targets on a shared host. Once deployed, the Gateway acts as a serverless ingress layer that accepts standard requests and directs them according to those rules. Read full analysis →

🧠 AirPods Pro 3 just got new firmware release in beta, more models too

Apple has released new beta firmware for AirPods Pro 3 and additional models. The update is expected to reach all users alongside iOS 27 next month, according to 9to5Mac. Read full analysis →

🛡️ Zenity Raises $125 Million in Series C Funding

Zenity has raised $125 million in Series C funding, according to SecurityWeek. The AI security company plans to put the capital toward product innovation, global expansion, and customer experience. Read full analysis →

🛡️ Obsidian Security Raises $85 Million at $1.1 Billion Valuation

Obsidian Security has raised $85 million at a $1.1 billion valuation. The company has developed a platform for governing AI agents across third-party applications, according to SecurityWeek. Read full analysis →

🛡️ Third-party cyber evaluations involving OpenAI models

OpenAI has addressed recent third-party cybersecurity evaluation incidents involving its models. In a statement on OpenAI News, the company explains those incidents and outlines new safeguards aimed at strengthening AI model testing and evaluation. Read full analysis →

🛡️ Django security releases issued: 6.0.8 and 5.2.17

The Django team has issued security releases for Django 6.0.8 and Django 5.2.17 under its security release policy. The updates address the security issues detailed in the announcement. All Django users are encouraged to upgrade as soon as possible. Read full analysis →

📰 Claude, GPT and the Minnesota attackers all used known bugs and weak logins

Claude, GPT, and the Minnesota attackers all used known bugs and weak logins, according to a piece on Cabreza’s Substack. The article is up on Hacker News with one point and no comments so far. Read full analysis →

🧠 Announcing the AI Glasses Impact Grant Recipients: Helping People Work, Learn, and Live More Independently

Meta Newsroom reports that the company has awarded AI Glasses Impact Grants to 30 organizations across the United States. The grantees are using Meta AI glasses to help people work more safely, learn faster, and live more independently. Read full analysis →

🧠 GH-ESD: Grounded Hypothesis-Driven Error Slice Discovery for Instance-Level Vision Tasks

Apple Machine Learning Research introduces GH-ESD, or Grounded Hypothesis-Driven Error Slice Discovery, for finding systematic failures of vision models on semantically coherent subsets known as error slices. Existing slice discovery methods mostly treat slices as clusters in representation space or combinations of predefined attributes, which can work for image-level classification but fall short for instance-level tasks such as object detection and segmentation. For those tasks, failures often stem from contextual, relational, and spatially grounded factors that earlier formulations do not adequately capture. Read full analysis →

🧠 Ling 3.0 Flash is now available on AI Gateway

Ling 3.0 Flash from Ant Group is now available on Vercel’s AI Gateway and is free to use for the next three weeks, through August 3rd. The Mixture-of-Experts model has 124B total parameters with about 5.1B active per token, a 256K token context window, and both thinking and non-thinking modes. It is built for token-efficient agentic inference at production scale. Read full analysis →

🔎 More Interesting News

• Andrew Ng made an open source agent — HN AI Agents
• BTL-3: A 27B open-weight agent model for agentic coding and structural tool use — HN AI Agents
• Codex Slides: open-source AI slide studio powered by Codex. Prompt, repo to deck — HN Claude/Codex/Fable
• Show HN: Deep Skill Finder – Find agent skills using real execution benchmarks — HN AI Agents
• Launch of the AI Infra SIG under the CNCF Japan chapter: First meetup and call for speakers — CNCF Blog
• WhatsApp launches a new design for message bubbles on iOS — WABetaInfo
• Handwritten-edit benchmark: Fable 5 is #1, Opus 4.8 regresses 55% on miscounting — HN Claude/Codex/Fable
• Save Up to 70% on Claude Code Costs with PNG Prompts? — HN Claude/Codex/Fable
• Australian energy provider Origin says data breach exposes client data — BleepingComputer
• DingDuff: A Claude MCP connector for legal research — HN Claude/Codex/Fable
• Minimize idle accelerators: Native RL job interleaving with co-operative time-slicing in llm-d — Google Cloud Blog
• Building trade assistant: How Jefferies optimized front office trading operations with AI — AWS Machine Learning Blog
• Meta launched a new AI optimism ad set to a song about human extinction — TechCrunch
• What the New 100x Agentic Engineer Looks Like in the Era of Fable and GPT 5.6 — HN Claude/Codex/Fable
• Training Agent Harness Like Training a ML Model — HN AI Agents
• The credential that let OpenAI's agents into Hugging Face exists in most enterprises right now — VentureBeat
• Running Claude Code on a Jailbroken Kindle over SSH via Tailscale — HN Claude/Codex/Fable
• UseReserve – a free macOS menu bar app for tracking Claude and Codex usage — HN Claude/Codex/Fable
• Travis Kalanick’s robotics company raises $1.7B, led by a16z — TechCrunch
• Menlo Ventures’ Matt Murphy explains why Anthropic is winning (and it’s not the model) — TechCrunch
• Samsung Unpacked 2026: all the news from the July foldable launch — The Verge
• eslint-rspack-plugin 5.0.0 Ships as a Pure ESM Package, Aligning with the Rstack Ecosystem — InfoQ
• Who’s legally to blame for Anthropic and OpenAI’s autonomous AI hacks? It’s complicated — TechCrunch
• DesignArena creators raise $7.9 million to bring taste to AI models — TechCrunch
• Hermes Agent v0.20.0 — HN AI Agents