# forck.live — 22–28 July 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-W30, 22–28 July 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-W30 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Vercel — Flags as code in Next.js - Company: Vercel (vercel.com) - Announced: 2024-07-26T13:00:00+00:00 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://vercel.com/blog/flags-as-code-in-next-js - Record: https://forck.live/items/20965-flags-as-code-in-next-js - Subject: v0 / AI SDK / AI Gateway Vercel introduced a Flags SDK for Next.js and SvelteKit that treats feature flags as server-side functions, enabling precomputed static pages and provider-agnostic flag management. ### Aleph Alpha — Lighthouse for sovereign AI development in Germany: Aleph Alpha strengthens collaboration with the IPAI ecosystem for open and application-oriented research and development in GenAI - Company: Aleph Alpha (aleph-alpha.com) - Announced: 2024-07-25 - Category: funding-company-news - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://aleph-alpha.com/en/news/ipai-collaboration-strengthened - Record: https://forck.live/items/16865-lighthouse-for-sovereign-ai-development-in-germany-aleph-alpha-strengthens - Subject: Kolibri / PhariaAI Aleph Alpha has founded Aleph Alpha Research, a company dedicated to applied AI research, with 300 million euros in long-term funding. The unit is owned by the non-profit IPAI Foundation and will make developed models, algorithms, and training code freely accessible for research and educational purposes. The research focuses on large language models, data- and compute-efficient adaptation, and traceability of AI models. ### Hugging Face — LAVE: Zero-shot VQA Evaluation on Docmatix with LLMs - Do We Still Need Fine-Tuning? - Company: Hugging Face (huggingface.co) - Announced: 2024-07-25 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/zero-shot-vqa-docmatix - Record: https://forck.live/items/1848-lave-zero-shot-vqa-evaluation-on-docmatix-with-llms-do-we-still-need-fine-tuning - Subject: Platform Hugging Face researchers introduce LAVE, an LLM-assisted evaluation metric for zero-shot VQA on the Docmatix dataset, questioning whether fine-tuning is necessary for accurate evaluation. ### OpenAI — SearchGPT is a prototype of new AI search features - Company: OpenAI (openai.com) - Announced: 2024-07-25 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/searchgpt-prototype - Record: https://forck.live/items/730-searchgpt-is-a-prototype-of-new-ai-search-features - Subject: GPT / ChatGPT / API OpenAI announces it is testing SearchGPT, a temporary prototype of new search features that provide fast and timely answers along with clear and relevant sources. ### Vercel — Elkjøp's Digital Transformation: Powering Retail Innovation with Next.js and Vercel - Company: Vercel (vercel.com) - Announced: 2024-07-24T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/elkjops-digital-transformation-with-next-js-and-vercel - Record: https://forck.live/items/20966-elkj-p-s-digital-transformation-powering-retail-innovation-with-next-js-and - Subject: v0 / AI SDK / AI Gateway Elkjøp, a Nordic consumer electronics retailer, migrated its ecommerce platform from Angular and Kubernetes to Next.js and Vercel. The transition reduced release cycles from 7 weeks to 1 day, improved Core Web Vitals (93% CLS improvement), and supported over $1 billion in digital revenue. ### Mistral AI — Large Enough - Company: Mistral AI (mistral.ai) - Announced: 2024-07-24T12:00:00+00:00 - Category: model-update - Coverage: not counted - Announcement: yes - Group: models - Source: https://mistral.ai/news/mistral-large-2407/ - Record: https://forck.live/items/7366-large-enough - Subject: Mistral / Le Chat Mistral AI released Mistral Large 2, a 123 billion parameter model with a 128k context window, supporting dozens of languages and over 80 coding languages. The model is available on la Plateforme and its weights are released under the Mistral Research License. Mistral Large 2 outperforms the previous Mistral Large and performs on par with leading models like GPT-4o, Claude 3 Opus, and Llama 3 405B. ### OpenAI — Improving Model Safety Behavior with Rule-Based Rewards - Company: OpenAI (openai.com) - Announced: 2024-07-24T09:00:00+00:00 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/improving-model-safety-behavior-with-rule-based-rewards - Record: https://forck.live/items/731-improving-model-safety-behavior-with-rule-based-rewards - Subject: GPT / ChatGPT / API OpenAI developed a new method called Rule-Based Rewards (RBRs) to align models for safe behavior without needing extensive human data collection. ### Harvey — Harvey Raises Series C from Google Ventures, OpenAI, Kleiner Perkins, Sequoia Capital, Elad Gil, and SV Angel - Company: Harvey (harvey.ai) - Announced: 2024-07-23T21:22:00+00:00 - Category: funding-company-news - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.harvey.ai/blog/harvey-raises-series-c - Record: https://forck.live/items/17732-harvey-raises-series-c-from-google-ventures-openai-kleiner-perkins-sequoia - Subject: Harvey Harvey announced a $100 million Series C funding round led by GV (Google Ventures) with participation from OpenAI, Kleiner Perkins, Sequoia Capital, Elad Gil, and SV Angel, at a $1.5 billion valuation. The company said it has tripled ARR and doubled employee headcount since its last fundraise in December, and that it will use the capital to invest in engineering, data, and domain expertise, as well as deepen partnerships with cloud and model providers. ### Vercel — How to choose the best rendering strategy for your app - Company: Vercel (vercel.com) - Announced: 2024-07-23T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/how-to-choose-the-best-rendering-strategy-for-your-app - Record: https://forck.live/items/20968-how-to-choose-the-best-rendering-strategy-for-your-app - Subject: v0 / AI SDK / AI Gateway Vercel's guide explains the use cases and tradeoffs of Static Site Generation (SSG), Incremental Static Regeneration (ISR), Server-Side Rendering (SSR), Client-Side Rendering (CSR), and experimental Partial Prerendering (PPR) for optimizing performance, SEO, and user experience. It covers ideal use cases, benefits, and implementation considerations for each strategy, and notes that the Next.js App Router defaults to streaming SSR. ### Vercel — Turbopack updates: Moving homes - Company: Vercel (vercel.com) - Announced: 2024-07-23T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/turbopack-moving-homes - Record: https://forck.live/items/20967-turbopack-updates-moving-homes - Subject: v0 / AI SDK / AI Gateway Vercel announced it will move the Turbopack bundler source code from the vercel/turbo repository into the vercel/next.js monorepo to accelerate development and achieve a stable release with production build support. The migration will use git-subtree and git-filter-branch to preserve commit history. Turborepo remains in its current repository, and Turbopack's core remains framework-agnostic despite the immediate focus on Next.js. ### Replicate — Run Meta Llama 3.1 405B with an API - Company: Replicate (replicate.com) - Announced: 2024-07-23 - Category: developer-tool-release - Coverage: not counted - Announcement: no - Group: routine - Source: https://replicate.com/blog/run-llama-3-1-with-an-api - Record: https://forck.live/items/2524-run-meta-llama-3-1-405b-with-an-api - Subject: Platform Replicate announces API support for running Meta's Llama 3.1 405B model, including usage examples in JavaScript, Python, and cURL, as well as details on the model's parameters and context window. ### Hugging Face — Llama 3.1 - 405B, 70B & 8B with multilinguality and long context - Company: Hugging Face (huggingface.co) - Announced: 2024-07-23 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://huggingface.co/blog/llama31 - Record: https://forck.live/items/1849-llama-3-1-405b-70b-8b-with-multilinguality-and-long-context - Subject: Platform Meta released Llama 3.1 models in three sizes (8B, 70B, 405B) with base and instruct variants, featuring 128K context length, multilingual support, tool calling, and a more permissive license. Also released Llama Guard 3 and Prompt Guard. Hugging Face integrates them into its ecosystem. ### Harvey — Harvey partners with Voyage to build custom legal embeddings - Company: Harvey (harvey.ai) - Announced: 2024-07-22T21:25:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.harvey.ai/blog/harvey-partners-with-voyage-to-build-custom-legal-embeddings - Record: https://forck.live/items/17733-harvey-partners-with-voyage-to-build-custom-legal-embeddings - Subject: Harvey Harvey partnered with Voyage AI to fine-tune a custom embedding model, voyage-law-2-harvey, on over 20 billion tokens of US case law. The model reduces irrelevant top results by nearly 25% compared to leading off-the-shelf embedding models while using one-third the embedding dimensionality, improving retrieval quality for legal queries. ### Hugging Face — WWDC 24: Running Mistral 7B with Core ML - Company: Hugging Face (huggingface.co) - Announced: 2024-07-22 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/mistral-coreml - Record: https://forck.live/items/1850-wwdc-24-running-mistral-7b-with-core-ml - Subject: Platform To run Mistral 7B with Core ML on a Mac, using new features from WWDC 24 such as Swift Tensor and Stateful Buffers, and achieving less than 4GB memory usage.