|
|
Hi there, this is your daily ☕️ DevShot.
|
|
|
In today's DevShot:
|
|
🤖 Tencent's Hy4 AI can operate computers 🧊 Google previews Three-Live for native 3D 🔒 GitHub blocks pull_request_target in public repos Dec 2 🧑💻 New 65B model tops US open rivals ⚡ Spark 4.3 adds real-time stateful processing Plus: 🎁 7 more stories you might like, 🛠️ 5 tutorials, 🧰 7 tools, and 📚 5 papers.
|
|
100+ ChatGPT Prompts to Revolutionize Your DayIndustry-Specific Use Cases: 15+ real-world applications across various sectors Productivity Guide: 21 best practices to 10x your efficiency with AI Prompt Powerhouse: 100+ ready-to-use prompts for immediate implementation Challenge Buster: Overcome common AI hurdles with expert strategies
Plus, in-depth sections on email composition, content creation, customer support, and data analysis.
|
🤖 Tencent's Hy4 AI can operate computers
LINK
|
- Tencent launched Hy4 today, a large language model that for the first time can operate computers directly, pairing visual understanding, reasoning and tool-calling to carry out long-horizon agentic tasks across real-world workplace and professional workflows.
- Built on 770 billion parameters with 49 billion active, it upgrades the August Preview by curbing excessive reasoning and over-verification, and integrates with Unreal Engine, Unity and Godot for complex game logic and multiplayer implementations.
- Pricing stays at roughly $0.90 per million input tokens and $2.70 per million output, access runs through Tencent Cloud TokenHub plus OpenRouter and OpenCode, and Tencent said it will release model weights and system cards in the coming days.
|
🧊 Google previews Three-Live for native 3D
LINK
|
- Google has released Three-Live, an open-source library that compiles LLM-generated Three.js code on-device and renders it natively using Filament and Scene Viewer on Android, RealityKit and Quick Look on iOS, and WebGL2 and WebXR on Web.
- Rather than relying on browser WebXR, which can't enter AR in iOS Safari or inside embedded native app views, Three-Live runs the same Three.js script on each platform's own engine, letting an agent produce a complete interactive scene in 15 to 30 lines of JavaScript with no external binary assets.
- The library, currently in Developer Preview (Alpha), executes client scripts inside a Web Worker or opaque-origin on the web and a sandboxed Rust VM on mobile with network and storage globals removed, CPU fuel limits, and memory quotas, and ships as prebuilt bundles on GitHub Releases.
|
🔒 GitHub blocks pull_request_target in public repos Dec 2
LINK
|
- GitHub Actions will block the `pull_request_target` event by default in all public repositories starting December 2, 2026, unless an Actions event policy explicitly permits it, enforcement slipped from the earlier November 2 date.
- The trigger runs in the base repository's context with access to its secrets, so untrusted fork code could steal them; the rule starts in evaluate mode, flagging which runs it would block before enforcement.
- Two brownouts, the first on November 9, will temporarily enforce the block and create failed runs with error annotations, letting you use Insights to spot affected workflows; private and internal repos and GitHub Enterprise Server are exempt.
|
🧑💻 New 65B model tops US open rivals
LINK
|
- Ricursive, a chip-design startup, released Ripple-65B Preview, an open-weights coding model it claims tops other U.S. open models on the Terminal-Bench 2.1 and SWE-bench Pro v2.0 Hard benchmarks.
- Trained from scratch on 12 trillion tokens with a dense architecture, Ripple narrowly edged Reflection's Beam on Terminal-Bench using about 15x fewer parameters than Thinking Machines' Inkling, with training reportedly a tenth of rivals' cost.
- It's on Hugging Face under Apache 2.0 with OpenRouter access at $0.20/$0.60 per million tokens, though the advertised million-token window is capped far lower there, and Ricursive's own figures show Ripple trailing rivals on several other benchmarks.
|
⚡ Spark 4.3 adds real-time stateful processing
LINK
|
- Apache Spark 4.3 extends Real-Time Mode to stateful processing, letting streaming aggregations, deduplication, and custom stateful logic run in Structured Streaming's low-latency execution path at sub-150 ms P99 end-to-end latency across three tested benchmarks.
- Supporting this required native concurrent stage scheduling and pipelined shuffles, so producer and consumer stages run together via a new Streaming Shuffle that transfers records incrementally instead of waiting for the entire producing stage to finish.
- For supported workloads you no longer need a separate streaming engine to hit millisecond-scale latency, using familiar Structured Streaming APIs in the same engine as batch and analytics, though the figures are measurements, not a latency guarantee for every pipeline.
|
|
Turn your expertise into working softwareDescribe what you want to build Get real software, not a prototype Lovable handles payments, logins, databases Change any detail yourself, any time Put it in front of customers this week
Discount of $5 applies to first month of any paid plan. You need to sign up via this webpage in order to claim the discount. The discount applies to new and returning users without an active paid plan. Terms apply.
📚 Tutorials & Advice
|
> Two dimensional version control: Using Jujutsu, a version control tool that auto-propagates edits across branches, lets you edit any tutorial chapter and have changes roll into the others automatically.
|
|
|
🔗 Other News & Articles
-
C++26 DIS failed, how does C++ survive?
LINK
-
vercel-labs/agent-desktop: Computer use CLI for AI agents
LINK
-
floci-io/floci: Light, fluffy, and always free - The AWS Local Emulator alternative
LINK
-
Could a modern chess engine on a Pentium have defeated Kasparov in 1996?
LINK
-
OpenAI to acquire developer tooling startup Pierre Computer
LINK
-
Jailbreaking my car's navigation system just so I can change the time
LINK
|
|
📄 Papers & Resources
|
> AI Express bundles text, image, and code tools into one pay-as-you-go platform, replacing multiple separate AI subscriptions with a single interface that reviews code, turns text into working websites, and checks grammar without technical skills.
LINK
|
|
> Game character AI maps 42 studies on language-model-powered video game characters, finding most give them short-term memory but few remember long-term, with response delays of ~3 to 15 seconds blocking commercial use.
LINK
|
|
> AI tool retrieval now pulls not just the obvious matching tool but the prerequisite tools a task secretly needs, improving accuracy by up to ~8 points with no training or manual setup required.
LINK
|
|
> Automated code migrations cleared Bloomberg's backlog of low-priority infrastructure upgrades across thousands of repositories, turning 2,333 AI-generated pull requests into 729 completed migrations, with human review, not generation, proving the real bottleneck.
LINK
|
|
> A multi-model review panel analyzes thousands of open-ended stakeholder comments at scale for about 14 to 34 per 100 threads, matching human coders closely while keeping teams in charge of the final interpretation.
LINK
|
|
|
|
🧰 Latest Tools
| ElevenLabs: turns any script into studio-quality speech in seconds, with 5,000+ voices across 70+ languages. Free to start. GENERATE FOR FREE |
|
tiun: an AI-native backend bundling authentication, payments, customer data, and analytics in one platform, so teams avoid stitching separate services together.
LINK
|
|
Monospace from Directus: wraps any SQL database with instant REST and GraphQL APIs plus a no-code app for collaborative data management.
LINK
|
|
iFixAi: audits AI agents with 250 inspections across 69 misalignment categories, flagging failures and giving engineers evidence to investigate and fix issues
LINK
|
|
Floot MCP: connects Claude and ChatGPT to build full-stack apps from chat, handling databases, logins, hosting, emails, and publishing automatically.
LINK
|
|
Kombai Gallery: browse hundreds of categories of web and mobile UI designs for inspiration, then remix them or hand them to your coding agent.
LINK
|
|
Web Search Agents by Nimble: domain-specific crawling agents that self-learn your use case to pull deeper research for company enrichment, regulations, and monitoring tasks.
LINK
|
|
|
|
💬 How did you find today's edition?
We read every reply — just reply to this email and let us know how we can improve!
|
|
|
