Google LLM router ➡️, Cloudflare Wallets 💳, Anthropic and Volta 🤝

Model routing on the Google Cloud API Gateway is now available in public preview. The API gateway provides a lightweight, serverless ingress layer ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ 

Sign Up |Advertise|View Online TLDR Together With TLDR AI 2026-08-05 IBM Bob makes workflows reviewable, testable and scalable (Sponsor) The value of IBM Bob, an AI development partner, isn't simply that it writes code. It also helps turn messy discovery work into artifacts engineers can review, test and reuse. For developers, Bob helps reduce discovery work time by analyzing systems, generating work that fits team conventions and supporting the software development lifecycle. CrushBank developers used IBM Bob to reason through architecture, move into implementation and inspected generated changes without leaving the workflow. Bob helped identify patterns, build ingestion paths and create solutions while keeping developers in control through code reviews, sensitive data scanning, test harnesses and human peer review. Read the full story 🚀 Headlines & Launches Anthropic Reportedly Signed a $10B Cloud Deal with Volta (3 minute read) Anthropic reportedly agreed to buy six years of cloud capacity from AI infrastructure startup Volta. The planned 133-megawatt Norway data center would be developed with Bitdeer and powered by NVIDIA Vera Rubin systems. A unified API for AI model routing (3 minute read) Model routing on the Google Cloud API Gateway is now available in public preview. The API gateway provides a lightweight, serverless ingress layer that accepts OpenAI-compatible requests and dynamically routes them to Gemini, Claude, or OpenAI OSS-GPT. It can be used standalone for simple rate limiting and token tracking or paired seamlessly with the Gemini Enterprise Agent Platform. This article contains a step-by-step guide on how to configure router logic. Cloudflare Introduced Programmable Wallets for AI Agents (4 minute read) Cloudflare Wallets is a system designed to give AI agents stable identities and controlled access to payments for APIs, MCP tools, and online content. Virtual Wallets will support spending limits, allow lists, and transaction caps for safer agentic commerce. 🧠 Deep Dives & Analysis Unpacking ChatGPT Work: the Agent for a Billion Users (18 minute read) ChatGPT Work is OpenAI's agent product for knowledge work. Its current form is an amalgamation of ChatGPT, the Codex app, the Codex harness, the original Codex cloud agent, ChatGPT agent, Atlas, OpenClaw, and more. This article decodes the complex product lineup and explains what Work is, where it fits in OpenAI's lineup, the many interesting choices in its design, the tensions underneath, and where the product is likely headed. OpenAI plans to merge Chat and Work, so Work is really a preview for how ChatGPT's billions of users will soon use the app. What Codex Actually Sends to the Model (10 minute read) A developer recorded the requests Codex generated for a 16-character prompt by pointing it at a custom local server. They measured what changed as it loaded instructions, exposed tools, read files, ran commands, received images, and compacted its history. The experiment did not call an external model. This post details the findings from the experiment. 🧑‍💻 Engineering & Research Black Duck: AI-driven exploits are here. ARE YOU READY? (Sponsor) The exploit window is collapsing. AI turns newly disclosed vulnerabilities into working attacks in hours, not weeks. Black Duck Polaris™ Platform and Signal™ help organizations become Mythos Ready with intelligent prioritization, automated workflows, and faster remediation of exploitable risk.  Black Duck Can Help. Introducing Shieldstral (5 minute read) Shieldstral features a 3B open-weights multimodal safety classifier that outperforms models up to 7x its size. It accepts plain-language policies at inference time, unifying text and image safety evaluation without retraining. Shieldstral delivers calibrated safety scores across diverse benchmarks while running efficiently on a single 16GB NVIDIA GPU. It is a step toward moderation that adapts to context instead of forcing every product through one frozen taxonomy. NVIDIA Released Alpamayo 2 (4 minute read) NVIDIA has released Alpamayo 2 Super under a commercial license for robotaxis and autonomous vehicles. The reasoning model was designed to handle rare driving scenarios with inspectable decisions and broad multitask capabilities. Introducing Kiro Crew (12 minute read) Kiro Crew is a persistent workspace for development work that self-improves and continues beyond one session. It can run locally or remotely on personal hardware. Kiro Crew allows developers to start work from within a desktop app, web dashboard, or CLI, and continue the same work through connection tools like Slack and Discord. It can run multistep tasks and recurring jobs on a schedule and also monitor systems until something needs attention. DiffusionGemma Technical Report (49 minute read) DiffusionGemma adapted Gemma 4 into a discrete diffusion model that refined 256-token blocks in parallel, reaching roughly 1,500 output tokens per second on a single NVIDIA H100. 🎁 Miscellaneous The Computer Use Verification Skill That Every Agent Needs (7 minute read) Computer-use verification lets coding agents reproduce bugs, validate implementations against specs, and attach screenshots or videos to pull requests. Integrated into triage, implementation, and review workflows, cloud-based subagents reduce human review burden and enable iterative self-debugging. SpaceX Says Spending Spree Is Supercharging AI Revenues (7 minute read) SpaceX's capital expenditures hit $18.4 billion in the most recent quarter. The bulk of that figure is tied to the company's ongoing AI build-out. The company is plowing money into terrestrial computing infrastructure, signing AI deals, and working toward launching orbital data centers. SpaceX is on track to have $100 billion in annualized recurring revenue by December, most of it coming from data center deals. ⚡ Quick Links Running Claude on enterprise data? Your token costs are adding up fast. (Sponsor) The cost of enterprise AI scales with every query. CData Connect AI can cut LLM context handling costs by up to 97.6% without sacrificing answers. Learn more about token-efficient architecture NVIDIA's Real-Time Full-Duplex Voice Model (6 minute read) NemotronLabs VoiceChat is an 11B end-to-end speech model that handles streaming understanding, speech generation, and tool calling within one architecture. The Reverse Replicator (6 minute read) Backflip AI developed a model that converts physical parts into digital CAD files in minutes for around $10. Mixture-of-Kittens: our open-source MoE megakernel for NVL72s (25 minute read) Cursor's Mixture-of-Kittens (MoK) is an open-source optimized Mixture-of-Experts (MoE) megakernel that improves efficiency on NVL72s GPUs, which significantly boosts performance for models like Composer by addressing computation and communication bottlenecks. LFM2.5-2.6B: Deploy Agents Everywhere (8 minute read) LFM2.5-2.6B, a 2.6B parameter on-device agentic model, enables free inference, low latency, and robust privacy by running locally on hardware like phones or CPUs. Love TLDR? Tell your friends and get rewards! Share your referral link below with friends to get free TLDR swag! https://refer.tldr.tech/0b6a6dc1/2 Track your referrals here. Want to advertise in TLDR? 📰 If your company is interested in reaching an audience of AI professionals and decision makers, you may want to advertise with us. Want to work at TLDR? 💼 Apply here, create your own role or send a friend's resume to jobs@tldr.tech and get $1k if we hire them! TLDR is one of Inc.'s Best Bootstrapped businesses of 2025. If you have any comments or feedback, just respond to this email! Thanks for reading, Andrew Tan, Ali Aminian, & Jacob Turner Manage your subscriptions to our other newsletters on tech, startups, and programming. Or if TLDR AI isn't for you, please unsubscribe.