Data Points: Anthropic updates Sonnet, Claude Code, and agentic dev tools

In today’s edition of Data Points, you’ll learn more about:

Two new Gemini robotics that coordinate to learn and act
ChatGPT bringing news and research to you with Pulse
Spotify’s crackdown on AI song impersonation
Microsoft introduces “vibe working” to Office

But first:

Anthropic’s Claude Sonnet 4.5 brings major upgrades for coding

Claude Sonnet 4.5 now leads on coding benchmarks like SWE-bench Verified and OSWorld, reflecting improved software engineering and computer task performance. The upgrade also brings new features to Claude Code, including checkpoints to save your work, a refreshed terminal interface, a native VS Code extension, and advanced memory tools for the Claude API. Anthropic also introduced the Claude Agent SDK, letting developers build complex agents using the same infrastructure as Claude Code. Claude Sonnet 4.5 is available now via API at $3 per million tokens for input and $15 per million tokens for output, matching the price of Sonnet 4, with product updates rolled out to all users. (Anthropic)

Perplexity launches Search API with real-time web index for developers

The new API gives developers programmatic access to the same infrastructure that powers Perplexity’s public answer engine, including an index covering hundreds of billions of webpages. The API returns structured, fine-grained search results optimized for AI and traditional applications, and supports real-time updates with tens of thousands of index operations per second. Perplexity also unveiled an SDK, open-source evaluation framework, and developer tools to ease integration and testing. The Search API costs $5 per 1000 requests; further details and sign-up information are available through Perplexity’s developer portal. (Perplexity)

Google introduces two advanced robotics models

Google’s two new AI models, Gemini Robotics 1.5 and Gemini Robotics-ER 1.5, are designed to help robots better understand, plan, and act in complex real-world environments. Robotics-ER 1.5 manages high-level planning, tool use, and state-of-the-art spatial reasoning, while Robotics 1.5 translates these plans into actions and learns to transfer skills across different robot types. Both models work together, allowing robots to break down multi-step tasks, explain their reasoning, and improve transparency and safety. These models mark a step toward building general-purpose physical agents and address safety through updated evaluation benchmarks. Gemini Robotics-ER 1.5 is available now to developers via Google AI Studio, while Robotics 1.5 is limited to select partners. (Google)

ChatGPT Pulse is OpenAI’s latest proactive research tool

OpenAI released a preview of ChatGPT Pulse to Pro subscribers on mobile, offering daily personalized updates based on chat history, user feedback, and connected apps like Gmail and Google Calendar. Pulse synthesizes information overnight to deliver visual cards with recommendations, reminders, and progress updates, which users can curate and refine through direct feedback. Safety checks filter out harmful material, and integration with third-party apps remains optional and off by default. With Pulse, OpenAI says it is moving to build AI assistants that anticipate user needs rather than waiting for prompts, an emerging trend with implications for developers building context-aware systems. The Pulse preview is available now for Pro users, with plans to expand to Plus and then all users after further testing. (OpenAI)

Spotify rolls out new policies to curb impersonation and spam

Spotify said it had removed over 75 million spam tracks from its platform in the past twelve months as generative AI tools accelerated mass uploads and abuse. Last week, the company introduced new measures to protect artists and listeners from AI misuse, including a clarified vocal impersonation policy, a spam filter system, and industry-standard AI disclosures in track credits. The new impersonation policy bans AI-generated vocal clones without the artist’s consent and speeds up the process for artists to report fraudulent uploads and profile mismatches. Spotify’s music spam filter will identify and suppress tracks that use tactics like mass uploads, duplicates, or artificially short songs to game royalties, which reached $10 billion in 2024. These updates will roll out this fall, with AI credit disclosures developed in partnership with industry groups like DDEX and distributors including CD Baby, DistroKid, and EMPIRE. (Spotify)

Microsoft launches Copilot Agent Mode and Office Agent

Microsoft released Agent Mode in Excel and Word, along with Office Agent in Copilot chat, to automate document creation and research tasks using advanced AI models. Agent Mode in Excel now generates, validates, and iterates spreadsheets based on user prompts, while Agent Mode in Word enables interactive document creation through conversational exchanges with Copilot. Office Agent in Copilot chat produces PowerPoint presentations and Word documents by clarifying user intent, conducting web research, and generating polished output that users can further refine. Agent Mode in Excel and Word and Office Agent are available today for Frontier program participants and U.S.-based Microsoft 365 Copilot Personal or Family subscribers on the web, with desktop support and broader rollout coming soon. (Microsoft)

Want to know more about what matters in AI right now?

Read the latest issue of The Batch for in-depth analysis of news and research.

Last week, Andrew Ng talked about China's decision to prevent major tech companies from purchasing Nvidia chips, highlighting its advancements in semiconductor independence, and the potential consequences for U.S. dependence on Taiwan's chip production.

“If China gains independence from Taiwan manufacturing significantly faster than the U.S., this would leave the U.S. much more vulnerable to possible disruptions in Taiwan, whether through natural disasters or man-made events. If manufacturing in Taiwan is disrupted for any reason and Chinese companies end up accounting for a large fraction of global semiconductor manufacturing capabilities, that would also help China gain tremendous geopolitical influence.”

Read Andrew’s letter here.

Other top AI news and research stories covered in depth:

Google’s AP2 provides developers with new tools to build agentic payments, in a bid to transform digital transactions.
A recent study reveals that ChatGPT users are now more likely to be young, female, and seeking information, highlighting demographic shifts in AI use.
Gambling sites are deploying AI tools that predict wins and track bets for sports fans, marking a new era in sports betting.
Researchers have developed a new technique that auto-selects training examples to speed up fine-tuning, advancing the efficiency of reinforcement learning.

Subscribe to Data Points