Toolkit

OpenRouter lets you reach models from many AI companies through a single API (a programming interface that lets your code talk to a service), according to Simon Willison. It "handles fallbacks automatically and picks the most cost-effective option for each request."

That convenience has a catch, flagged by developer Mohamed Moustafa. One model is often served by several backend providers running different software, so identical requests can behave differently. Some providers lack vision support even when the model has it.

To start, create an account at openrouter.ai. Call its /endpoints method with a model ID to see which providers serve that model, then lock requests to the one you want with the provider.only option.

Worth knowing: Without pinning a provider, the same model can give different results between requests, which matters if you are testing or comparing models.

IBM has released Granite Time Series PatchTST-FM-r2, a forecasting model that predicts future values in any numerical time series — sales figures, energy loads, website traffic, sensor readings — without needing to be trained on your specific data first. That approach is called zero-shot forecasting (generating predictions from a pre-trained model, with no fine-tuning required on new data).

On the GIFT-Eval leaderboard, a benchmark for time series forecasting run by independent researchers, the model ranks second overall among comparable zero-shot models. It is first among models carrying a permissive commercial licence — specifically Apache 2.0 (a licence that lets you use, modify, and deploy the model in commercial products without royalty fees).

The model weights, code, and benchmark reproduction steps are all available on Hugging Face today: Granite Time Series PatchTST-FM-r2.

Worth knowing: The model requires approximately 385 million parameters to run — check that your available hardware can handle that before committing to it for production use.

AlphaGenome Atlas is a free research tool from Google DeepMind that maps the predicted effect of every possible single-letter change in the human genome — roughly 9 billion variants — in one searchable database. It is available today through a browser-based portal, no coding required.

The tool is built on AlphaGenome, an AI model trained to predict how genetic changes affect biological processes like gene regulation. The Atlas precomputes those predictions at scale so researchers can look up any variant without running the model themselves. A companion score, the AVI (AlphaGenome Variant Impact) score, condenses the predictions into a single number to help rank variants quickly.

Early collaborators have already used it to find key variants in unsolved rare disease cases. The dataset is 1 petabyte — too large to download whole — but the web portal makes it accessible to researchers without computational infrastructure.

You can access it today at deepmind.google/blog/alphagenome-atlas.

Worth knowing: This is a research tool for scientists working in genomics and molecular biology — it requires domain knowledge to interpret results. General readers will find the portal interesting, but the outputs are not designed to be self-explanatory.

Blender is a free, professional-grade 3D modelling and rendering application. Frontier AI models have become genuinely capable at writing Blender's Python scripting language, which means you can describe a scene in plain English and have a coding agent build and render it for you — no 3D modelling experience needed.

Developer Simon Willison has been doing exactly this on his Mac using ChatGPT's Codex (a code-writing assistant built into ChatGPT, available with Codex access on your account), detailed at simonwillison.net.

To try it: download Blender from blender.org and install it at the default path. Then give the agent a prompt like:

"Use the already installed /Applications/Blender to render a scene of a pelican riding a bicycle."

Follow up with plain English refinements — "add a background," "make it more detailed" — and the agent handles the code. The output is a real .blend file you can open and edit yourself.

Worth knowing: Subscription requirements for Codex access were not confirmed at time of writing — check your ChatGPT plan before assuming it is included.

Hugging Face has released @huggingface/kernels, a JavaScript library that loads and runs GPU-optimised operations for machine learning directly in the browser using WebGPU (a browser API that lets web pages use your device's graphics card for computation), as detailed in the Hugging Face blog post.

The release ships with an initial collection of 207 kernels — the individual mathematical operations, like matrix multiplication and normalisation, that underpin most AI models. Each is versioned, documented, and testable.

Also included: Fleet, a browser-based benchmarking tool that runs these kernels on your own hardware and contributes performance data back to the project.

Who this is for: JavaScript developers building browser-based AI features. You will need a modern browser with WebGPU support (Chrome 113+ or Edge 113+). No server, no API key, no GPU cluster required.

Get started at huggingface.co/webgpu-kernels. The library is Apache 2.0 licensed, meaning free for commercial use.

Worth knowing: Performance varies significantly by device and browser — Fleet exists precisely because behaviour across real-world hardware is still being mapped.

OpenAI and Thailand's Ministry of Higher Education, Science, Research and Innovation have launched an eight-week accelerator program for ten Thai startups working in health, wellness, and education. According to OpenAI's announcement, each team receives US$2,000 in API credits (access tokens that let developers call OpenAI's models from their own software), one-on-one technical mentoring, and structured sessions covering product design, safety, cost management, and fundraising.

This is not a tool you can open in a browser today. It is a structured program for early-stage founders who already have a working prototype. The ten selected startups are named — CARIVA, insKru, Globish and seven others — and cohort applications are not currently open.

If you are a founder in Southeast Asia building on AI, this is a model worth watching as a template for similar public-private programs.

Worth knowing: The program is invitation-only for this cohort. API credits are provided for development use — costs beyond the granted amount would be the startup's own.

Plaud One is a pair of earbuds designed to record, transcribe, and summarise your conversations. The earbuds can also be used through their standalone charging case, which includes built-in 4G connectivity so recordings can be processed without your phone or Wi-Fi.

Each earbud has three microphones and records up to six hours on a single charge. The case extends total battery life to 36 hours and has a five-metre recording range. An AI button on the case connects to a feature called Plaud Agent, which integrates with Gmail, Google Calendar, Notion, and Slack.

The Plaud One costs $249.99 and is available to preorder now, with shipping expected in late September. Details on which markets are supported are not specified. Read more or preorder at The Verge.

Worth knowing: Conversations are processed on Plaud's servers, not on-device — the source does not specify whether every recording is uploaded automatically or only on request, so treat anything sensitive accordingly.

LLM is a command-line tool that lets you run queries against AI models directly from your terminal, without opening a browser. The llm-anthropic plugin connects it to Anthropic's Claude models. Version 0.27, released on 24 August 2026, updates compatibility with the new Anthropic Python library (v1.0.0), which swaps its underlying HTTP client. If you use LLM with Claude and your requests recently broke, this is the fix.

To get it: install or upgrade via pip install llm-anthropic. You will need an Anthropic API key and Python on your machine.

Worth knowing: This is a compatibility patch, not a feature release — it restores existing functionality rather than adding new capabilities. If your current setup is working fine, there is no urgent reason to upgrade today.

llm-openrouter is a plugin for Simon Willison's llm command-line tool that lets you query any model available on OpenRouter (a service that provides access to a wide range of AI models through a single API) directly from your terminal.

Version 0.7 brings three practical additions: built-in server-side tools for shell commands, web fetching, and web search, enabled with flags like -T WebSearch. It also now displays reasoning traces (the step-by-step thinking some models produce before giving a final answer) for models that support them.

To use it, install the llm tool, then add the plugin. Enable web search on any query with -T WebSearch.

Best suited for developers or technically comfortable readers who want a fast, scriptable way to compare models or run quick tasks from the command line.

Worth knowing: OpenRouter charges per token used — costs add up quickly if you run large batches or use frontier models. Check your usage dashboard.

Replit is a browser-based platform that lets you build and run software without installing anything. Previously, using its AI-powered features cost credits. Replit has now launched Free Mode, powered by GPT‑5.6 Luna (an OpenAI model from the GPT‑5.6 series), which lets anyone plan, prototype, and explore software ideas without consuming a usage budget, according to OpenAI's announcement.

In practice: you describe what you want to build in plain language, and Replit's AI helps you shape and develop it. When a task needs heavier reasoning, it routes to a more capable model, then returns to Free Mode automatically.

To get started, go to replit.com, create a free account, and open a new project. Free Mode is available without any payment details.

Worth knowing: Free Mode covers exploration and planning — actually deploying or running your finished app moves you into Replit's paid Build Mode, so check the pricing before you get attached to a project.

LM Studio is a free desktop app that lets you download and run open-weights models (models whose trained files are publicly available for download) on your own machine, with no data sent to a third party.

Qwen 3.8 27B is available in LM Studio now. Search for the Q4_K_M quantized build — quantization (a compression technique that shrinks the model file while preserving most of its quality) brings it to a 17GB download. The source article notes this specific file size for the build tested.

Once loaded, the first thing to change is the reasoning effort setting. Go to model settings and drop it from "xhigh" to "low" or off entirely for everyday tasks. Then set the context window (the amount of text the model holds in memory at once) to at least 32,768 tokens, or higher if your machine allows.

To get started: download LM Studio, search for Qwen 3.8 27B, and select the Q4_K_M build.

Worth knowing: The 17GB file requires significant free disk space, and performance will depend heavily on your available RAM — check LM Studio's model card for your specific build before downloading.