OlmoEarth v1.3 releases global satellite embeddings and a browser-based mapping explorer
The OlmoEarth team released OlmoEarth v1.3 with precomputed 10-meter satellite embeddings in 64 and 128 dimensions, plus a browser-based Embeddings Explorer requiring no account or GPU. The team reports that its 128-dimensional embeddings lead on 20 of 27 evaluations against AlphaEarth Foundations and Tessera v2. Model weights, pretraining data, and code are open; at launch, embeddings cover the globe for 2023–2025 and select countries for 2017–2022.
Model releases and updates
The Verge · AIReported by 1 other sourcesCanva’s first AI image model plans designs before generating them
Canva launched Canva Craft, an image generation model whose two-stage architecture plans designs before generating images, which Canva says improves precision, consistency, and control. It supports up to 2K resolution and more than 30 aspect ratios, and Canva says it trained the model on millions of human-made templates. Craft is available now to all users globally through Dream Lab, with Craft-fast, Craft-ultrafast, and access through Model Context Protocol planned.
Product releases and updates
AWS Machine Learning BlogFirst-partyAmazon SageMaker training session APIs enable custom training loops on managed GPUs
AWS introduced Amazon SageMaker AI training session APIs, letting developers write custom Python training loops while the service manages GPU infrastructure, distributed execution, fault recovery, and checkpointing. Usage is billed per token processed, excluding local reward computation time. The accompanying cookbook provides SFT and RL recipes, including a competitive programming example where GPT-OSS 20B held-out accuracy rose from 40 percent to 50.5 percent on 200 test problems after 50 training steps.
The DecoderReported by 2 other sourcesAnthropic previews Claude Duet, a proactive agent that works while users are away
Anthropic introduced a research preview of Claude Duet, a standalone proactive agent built on Claude Opus 5.5 that works on tasks while users are away. Available through a waitlist for US Pro and Max subscribers on iOS, Android, and the web, it checks connected apps every few hours and shares usage limits with other Claude products. Anthropic says Duet requires approval before sending messages, making payments, signing documents, or deleting anything, and acknowledges it can misread threads or miss updates.
TechCrunch · AIOpenAI expands Decisions API availability as competition with TypeSafe’s Jev heats up
OpenAI made its Decisions API available to all customers on Thursday, adding voice and visual modalities and geographically targeting inference resources to reduce latency. The API uses a reconfigured ChatGPT Luna rather than a specially trained decision model; customer Speak reported cutting time between actions from 1.4 seconds to .13 seconds. Competing models ranked higher on two public decision model benchmarks.
The DecoderGoogle's Agent Compute Engine gives each AI agent a hardware-isolated microVM
Google announced Agent Compute Engine, which already powers Gemini and gives each AI agent an isolated microVM secured by Google's Titanium chip, with preview access coming in the coming months. Idle VMs pause automatically, while shared memory reduces operating costs. Google says its open-sourced Kubernetes runtime, Agent Substrate, supports ten times as many agents per server and achieved suspend and resume times below 500 milliseconds in a test with 200,000 agents.
Claude makes auto mode the default across apps for Pro, Max, and Team
Starting next week, sessions across Claude web, desktop, and mobile apps will run in auto mode by default on Pro, Max, and Team plans. Rather than asking you to approve each step, an automated check blocks risky actions. Switch back to manual anytime. https://twitter.com/ClaudeDevs/status/2085794862608318627
Industry
Anthropic donates Agent Skills to the Agentic AI Foundation
Agent Skills started as an experiment in Claude Code. Boris Cherny says industry adoption led Anthropic to donate it to the Agentic AI Foundation, alongside MCP. https://twitter.com/ClaudeDevs/status/2113328526413996100 In the quoted announcement, @ClaudeDevs says 70+ agent platforms have integrated the standard since December 2025, including ChatGPT, Cursor, GitHub Copilot, and VS Code.
Latent SpaceOpenAI’s Patch the Planet and Wiz report independent PHP2Shell discoveries
Latent Space reports that Calif researchers working under OpenAI’s Patch the Planet project and Wiz independently discovered two PHP-core vulnerabilities named PHP2Shell. Calif used GPT-5.6 Sol, while Wiz credited Gemini 4 Argon and GPT-5.6 Sol alongside Wiz Atlas. Researchers reported unauthenticated remote code execution against WordPress, Drupal, and Joomla. OpenAI describes approximately two months of coordinated remediation; Cloudflare deployed temporary WAF protections, while the report distinguishes demonstrated exploits from broader, uncorroborated exposure and outage claims.
Research
Google DeepMindFirst-partyGoogle DeepMind and partners release ForTy maps tracking forest types and change
Google DeepMind, Google Research, WRI and IIASA released Forest Typology (ForTy) 2020, a global 10-meter map classifying land into six forest and land-use categories, alongside an early time series covering 2017 to 2025. Validation across more than 9,000 sites found 90% accuracy for forest types and 95% for distinguishing forest from non-forest land. The datasets are freely available on Google Earth Engine under an open license and show changes including natural forests becoming plantations or farms.
Tips and perspectives
Turning Database Grading into RL Rewards: Post-Training Agents with ThinkingBox
Microsoft and Hugging Face released ThinkingBox-Training on GitHub, turning executable checks of final database states into rewards for agent post-training. After 50 GRPO steps on 187 tasks outside ThinkingBox-Bench, Qwen3.8-27B’s average benchmark success rose from 51.70% to 60.78%, and tasks passed on all 20 attempts increased from 38 to 107, while tasks solved at least once fell from 453 to 446. The release includes task selection, rollout, reward, and Verl integration code; the reported run’s private training pool is not included.
AWS Machine Learning BlogFirst-partyStream low-latency speech from AI agents using Kokoro TTS on Amazon SageMaker AI
AWS demonstrates how to stream completed sentences from an AI agent to Kokoro-82M on Amazon SageMaker AI so speech synthesis begins before the full response is generated. Its tests reduced mean time to first audio byte from 329.3 ms to 156.7 ms in isolation and measured 3.5 times faster first audio through the deployed AgentCore runtime. The guide covers persistent bidirectional sessions, binary WebSocket frames, and audio transport tradeoffs involving SSE and base64.
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 · 2026-10-22