# forck.live — 29 July – 4 August 2024 > AI lab announcements, news reports and podcasts. forck.live follows what companies announce about their own products. Press may corroborate an official story; it never leads and never appears as an item of its own. A missing date stays missing. A title is the publisher's words — where a company does not publish in English, an English rendering is listed separately as `Title in English` and named as this site's translation, never as theirs. This file is one issue of the weekly archive: every confirmed first-party announcement of ISO week 2024-W31, 29 July – 4 August 2024, newest first. The window never moves, so this URL always names the same seven days — but a post published inside them and detected later is added when it arrives, so a closed week can still gain entries. Every entry is one first-party announcement, and carries the same eight fields in this order before any others: Company · Announced · Category · Coverage · Announcement · Group · Source · Record `Source` is the company's own page the announcement was traced to and `Record` is this site's page for it. Both are safe to cite. `Coverage` is a number of outlets, or `not counted` — which means this record reached no count of its own, never that nobody wrote about it. `Category` is this tracker's own analysis, and is authoritative wherever it is stated. `not stated` means either that nothing has analysed the post yet or that the analysis placed it outside this taxonomy; neither is a verdict about the announcement. `Announcement` is whether the post is news of something the company did, rather than a how-to, a customer story, a changelog listing or a staffing note. `Group` is which of this record's groups the entry falls in, carried on the entry itself so it holds where the groups are headings and where they are not. One of: `models`, `covered`, `announcements`, `routine`. Anything after `Record` is held detail that only some announcements carry. - This week's page: https://forck.live/week/2024-W31 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Vercel — Protecting your app (and wallet) against malicious traffic - Company: Vercel (vercel.com) - Announced: 2024-08-02T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/protecting-your-app-and-wallet-against-malicious-traffic - Record: https://forck.live/items/20962-protecting-your-app-and-wallet-against-malicious-traffic - Subject: v0 / AI SDK / AI Gateway Vercel's guide covers how to block malicious traffic using custom Firewall rules, set spend limits, and apply code optimizations. It recommends enabling Fluid compute to save up to 85% on function usage and notes that Image Optimization billing changed from unique source images to transformations at a lowered price. ### Replicate — Replicate Intelligence #9 - Company: Replicate (replicate.com) - Announced: 2024-08-02 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replicate.com/blog/replicate-intelligence-2024-08-02 - Record: https://forck.live/items/2521-replicate-intelligence-9 - Subject: Platform Weekly bulletin covering new open-source AI models and tools, including FLUX.1 image generation, SAM 2 segmentation, Gemma 2 2B language model, and Gemma Scope interpretability tool. ### Replicate — FLUX.1: First Impressions - Company: Replicate (replicate.com) - Announced: 2024-08-02 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://replicate.com/blog/flux-first-impressions - Record: https://forck.live/items/2520-flux-1-first-impressions - Subject: Platform Replicate introduces FLUX.1, a new text-to-image AI model using flow matching instead of diffusion, and shares first impressions of its capabilities in text rendering, light and texture, artistic styles, composition, and a unique 'flow' aesthetic. ### Vercel — Achieving feature rollouts with ultra-low latency and zero impact to conversion - Company: Vercel (vercel.com) - Announced: 2024-08-01T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/beyond-menu-scaling-with-hypertune-and-vercel - Record: https://forck.live/items/20963-achieving-feature-rollouts-with-ultra-low-latency-and-zero-impact-to-conversion - Subject: v0 / AI SDK / AI Gateway Beyond Menu, a US food delivery service, adopted Hypertune and Vercel's Edge Config to manage feature flags and A/B testing in their Next.js app on Vercel. By syncing flag logic into Edge Config, they achieved SDK initialization on the server with ultra-low latency, avoiding layout shifts and UI flickers. The solution works across server and client, supports flexible targeting for anonymous visitors and restaurants, and integrates A/B tests directly into feature flag rules. ### Suno — The Future of Music - Company: Suno (suno.com) - Announced: 2024-08-01 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://suno.com/blog/future-of-music - Record: https://forck.live/items/10346-the-future-of-music - Subject: Music / Studio Suno responds to a copyright lawsuit filed by major record labels on June 24, arguing that training its music generation AI on copyrighted material from the open internet is not infringement, comparing it to a human learning musical styles. The company states it was in productive discussions with some RIAA members before the lawsuit and emphasizes its originality-guarding features. ### Replicate — Run FLUX with an API - Company: Replicate (replicate.com) - Announced: 2024-08-01 - Category: api-release - Coverage: not counted - Announcement: no - Group: routine - Source: https://replicate.com/blog/flux-state-of-the-art-image-generation - Record: https://forck.live/items/2522-run-flux-with-an-api - Subject: Platform Replicate announces that FLUX.1, a new open-source image generation model from Black Forest Labs, is now available on its platform via an API, with pricing per image and three variants. ### Vercel — How Google handles JavaScript throughout the indexing process - Company: Vercel (vercel.com) - Announced: 2024-07-31T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/how-google-handles-javascript-throughout-the-indexing-process - Record: https://forck.live/items/20964-how-google-handles-javascript-throughout-the-indexing-process - Subject: v0 / AI SDK / AI Gateway Vercel and MERJ published research analyzing over 37,000 Googlebot fetches to test common beliefs about JavaScript rendering and SEO. The study, conducted from April 1 to April 30, 2024, focused on nextjs.org and two other sites, using custom Edge Middleware to track rendering behavior. The article examines the evolution of Google's rendering capabilities and addresses myths such as 'Google can’t render JavaScript content.' ### Hugging Face — Google releases Gemma 2 2B, ShieldGemma and Gemma Scope - Company: Hugging Face (huggingface.co) - Announced: 2024-07-31 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://huggingface.co/blog/gemma-july-update - Record: https://forck.live/items/1845-google-releases-gemma-2-2b-shieldgemma-and-gemma-scope - Subject: Platform Google has released three new Gemma model variants: Gemma 2 2B (a 2.6B parameter small model), ShieldGemma (safety classifiers), and Gemma Scope (sparse autoencoders for Gemma 2 2B and 9B). ### Isomorphic Labs — Using our state-of-the-art AI models to power drug design - Company: Isomorphic Labs (isomorphiclabs.com) - Announced: 2024-07-30 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.isomorphiclabs.com/articles/using-bespoke-ai-models-to-power-drug-design - Record: https://forck.live/items/19145-using-our-state-of-the-art-ai-models-to-power-drug-design - Subject: Drug Design Engine / AlphaFold Isomorphic Labs describes how its multi-disciplinary drug design team uses proprietary AI models, including AlphaFold 3, to predict ligand-protein interactions and optimize drug candidates. The models are engineered for use by scientists without computational chemistry training, and the company aims to accelerate drug discovery by reducing the need for lab synthesis and testing. ### Hugging Face — Memory-efficient Diffusion Transformers with Quanto and Diffusers - Company: Hugging Face (huggingface.co) - Announced: 2024-07-30 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/quanto-diffusers - Record: https://forck.live/items/1846-memory-efficient-diffusion-transformers-with-quanto-and-diffusers - Subject: Platform Hugging Face announces that its Quanto quantization toolkit can be used with Diffusers to reduce GPU memory consumption for Transformer-based diffusion pipelines, including PixArt-Sigma, Stable Diffusion 3, and Aura Flow, with minimal quality loss. ### Nous Research — The Instruct Monomyth: why base models matter - Company: Nous Research (nousresearch.com) - Announced: 2024-07-29T21:52:10+00:00 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://nousresearch.com/the-instruct-monomyth - Record: https://forck.live/items/24038-the-instruct-monomyth-why-base-models-matter - Subject: Hermes models There is a deep, twisty labyrinth buried under a mountain of language, of symbol manipulation, and semantic nets. Its roots reach down deep into the Earth, absorbing the minutia of current thought, the limitations of logic, the constrained realm of rationality. ### Hugging Face — Serverless Inference with Hugging Face and NVIDIA NIM - Company: Hugging Face (huggingface.co) - Announced: 2024-07-29 - Category: infrastructure-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/inference-dgx-cloud - Record: https://forck.live/items/1847-serverless-inference-with-hugging-face-and-nvidia-nim - Subject: Platform Hugging Face and NVIDIA launch a serverless inference service called NVIDIA NIM API (serverless) on the Hugging Face Hub, available to Enterprise Hub organizations, providing pay-as-you-go access to open models like Llama and Mistral using NVIDIA DGX Cloud accelerated compute.