Clean data isn't a coding problem (Sponsor)
Say goodbye to the data bottleneck. When you can clean and organize data with no-code and AI-native tools, your humans and AI agents have the structured and searchable information they need to reason and act.
This Algolia whitepaper lays out the pillars of no-code data prep. You'll learn:
- Why clean data matters more than ever
- The pitfalls of traditional methods
- The key capabilities and benefits of modern no-code data tooling
Find the path through the data bottleneck.
Download the whitepaper
🚀Headlines & Launches
ChatGPT Health (3 minute read)
OpenAI launched Health in ChatGPT for US adults, allowing users to connect Apple Health and supported medical records for more contextual health conversations. The company said connected health data and related chats would not be used to train foundation models or target ads.
Announcing Fugu-Ultra v1.1 🐡 (1 minute read)
Fugu-Ultra v1.1 is more capable across coding, agentic tasks, and advanced reasoning. It is now available at the same price as Fugu-Ultra v1.0. Fugu offers frontier-level performance without single-vendor dependency. It dynamically orchestrates the best models to tackle complex multi-step tasks.
Runway Launched an AI Router for Generative Media (2 minute read)
Runway Media Router is a developer tool that automatically selected image, video, or audio models based on quality, speed, or cost. The release positions Runway Dev as an infrastructure layer offering unified API access to both Runway and third-party models.
Updated Claude Voice Mode (9 minute read)
Anthropic updated Claude's voice mode to support Opus, Sonnet, and Haiku, while adding integrations with apps such as Gmail, Slack, Notion, and Google Calendar. The beta remains available to all users, with free accounts limited to Haiku and one connected app.
🧠Deep Dives & Analysis
How to Build a Frontier Model Factory (48 minute read)
Poolside co-CEO Eiso Kant explains how a compact research team built the training infrastructure behind Laguna S, an 118B mixture-of-experts model reported to outperform much larger open-weight systems.
A GPU-Hour Isn't a Commodity If You Need Four of Them (6 minute read)
Compute pricing coverage runs on headline rates, but many workloads don't consume them that way. Training, fine-tuning, and high-throughput inference need multiple identical GPUs in the same machine or cluster. A company that buys H100 futures to hedge a future four- or eight-GPU requirement carries real basis risk. The futures will hedge a rise in the broad H100 price and do nothing about the lack of co-located capacity.
Kimi K3's Design Secret may be in its Thinking Traces (6 minute read)
Kimi K3 uses an extreme amount of thinking tokens. It uses over 12 times more reasoning than Claude Opus 4.8 and over double that of Kimi K2.6. The model's performance is primarily due to its unique chain-of-thought approach in which it iterates upon designs like a full AI agent would, but inside its chain of thought. This article looks at Kimi K3's thinking traces to explain the model's performance.
👨💻Engineering & Research
OpenWorker (GitHub Repo)
Runs a local desktop AI coworker that completes tasks across files and apps.
Microsoft's New MAI-Image and MAI-Voice (2 minute read)
Microsoft introduced MAI-Image-2.5-Pro for high-fidelity image generation and editing, alongside MAI-Voice-2-Flash for faster, lower-cost voice applications. Both models entered public preview and joined Microsoft's broader production model lineup.
FLUX 3 (5 minute read)
Black Forest Labs has launched FLUX 3, a multimodal model that can generate images and audio-video clips up to 20 seconds from a single prompt. Its architecture could be used in the future in robotic perception and action.
🎁Miscellaneous
TLDR is hiring a curator for TLDR Hardware! (TLDR Curator, ~3 hrs/week)
500,000 people have already signed up for TLDR Hardware, our new twice-weekly newsletter covering chips, robotics, energy, and devices. If you work in hardware and want to help curate it, send your LinkedIn or resume to
hardware@tldr.tech!
Intel's stock jumps as chipmaker rides AI boom to fastest revenue growth in almost 15 years (4 minute read)
Intel reported better-than-expected results for its second quarter. The company's shares are up over 170% so far in 2026. It is starting to craft long-term agreements with customers for its server CPUs, a move that's becoming common in the industry. The company's CFO says Intel is supply constrained, with data center customers demanding more than it can produce.
Understanding the AI economy (7 minute read)
Google Research discusses the AI economy's rapid growth, focusing on innovations and impacts across industries. The piece highlights AI's transformative effects on sectors like healthcare and finance. It emphasizes the need for strategic investment and regulation to harness AI's potential responsibly.
DeepSeek founder Liang Wenfeng in His Own Words (18 minute read)
This post contains 64 quotes from DeepSeek's investor call. In the call, founder Liang Wenfeng discussed the company's vision, motivation, and culture. He also detailed the company's open-source strategy, plans for commercialization, roadmap to AGI, and much more.
⚡️Quick Links
CoderPad's CEO says “AI fluency” is fake. The VP of Product disagrees. Watch them debate (Sponsor)
How should AI change hiring? On August 5, watch CoderPad's execs debate AI hiring's biggest assumptions. Spoiler alert: they don't see things the same way.
Save your spotWelcoming The Interaction Company (2 minute read)
Cognition has added The Interaction Company of California, the makers of Poke, a personal agent that can text natively on Apple Messages, to its team.
Sierra acquires TakeOff, the long-horizon AI agent platform (3 minute read)
The deal gives Sierra immediate access to a platform that can power autonomous, multistep applications across enterprises.
AMD and Cerebras Launch AI Inference Solution (10 minute read)
AMD and Cerebras have unveiled an AI inference solution that combines AMD Helios with Cerebras Wafer-Scale Engine, enhancing low-latency AI workloads.
Agentic coding goes hands-free as OpenAI brings GPT-Live's full duplex voice control to Codex and ChatGPT on the desktop (5 minute read)
OpenAI's naturalistic GPT-Live audio AI model is now integrated with the ChatGPT desktop application.
Progress (3 minute read)
Etched's new architecture enables AI chips to run math blocks at half the usual voltage, boosting FLOPs density and tackling thermal issues.
DeepSeek's Huawei-Chip Training Claim Finally Gets Its Benchmarks, and Its Doubters (3 minute read)
A Huawei-led consortium has released a technical report that documents full-parameter post-training of DeepSeek's V4 family on Huawei Ascend chips, measuring the run at 34.22% model FLOPs utilization.