# forck.live — 18–24 November 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-W47, 18–24 November 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-W47 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Harvey — Introducing the Next Version of Vault - Company: Harvey (harvey.ai) - Announced: 2024-11-21T21:28:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.harvey.ai/blog/introducing-the-next-version-of-vault - Record: https://forck.live/items/17719-introducing-the-next-version-of-vault - Subject: Harvey Harvey launched an upgraded version of Vault, its collaborative workspace for document review, increasing the file limit per project from 1,000 to 10,000 documents. The update introduces one-click workflows for common document types (e.g., merger agreements, stock purchase agreements, leases) that extract 25+ data points per document, with recall scores ranging from 96.49% to 99.66%. Vault is now integrated into Assistant, allowing users to query Vault documents without switching pages. ### Vercel — Life of a Vercel request: Navigating the Edge Network - Company: Vercel (vercel.com) - Announced: 2024-11-21T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://vercel.com/blog/life-of-a-vercel-request-navigating-the-edge-network - Record: https://forck.live/items/20921-life-of-a-vercel-request-navigating-the-edge-network - Subject: v0 / AI SDK / AI Gateway Unpacking the core usage metrics of Edge Requests and Fast Data Transfer Vercel’s framework-defined infrastructure provisions cloud resources while providing full transparency, from the initial build to every incoming request. ### OpenAI — Advancing red teaming with people and AI - Company: OpenAI (openai.com) - Announced: 2024-11-21T10:30:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/advancing-red-teaming-with-people-and-ai - Record: https://forck.live/items/669-advancing-red-teaming-with-people-and-ai - Subject: GPT / ChatGPT / API OpenAI announced advancements in red teaming using both human and AI approaches. ### LG AI Research — LG AI Insight 2024 On-Site Report: A Look at This Year’s Achievements and Future AI Trends - Company: LG AI Research (lgresearch.ai) - Announced: 2024-11-21 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/news/view?seq=501 - Record: https://forck.live/items/11017-lg-ai-insight-2024-on-site-report-a-look-at-this-year-s-achievements-and - Subject: EXAONE LG AI Research held its LG AI Insight 2024 conference on November 21st, recapping its past-year achievements and discussing future AI trends. The event featured expert talks on AI agents, large language models, multimodal AI, bio and medical AI, and computing infrastructure, plus a poster session and demos of its EXAONE and ChatEXAONE services. The company said it will continue research on applying AI technologies, including EXAONE and ChatEXAONE, in the industrial field. ### LG AI Research — LG AI Insight 2024 On-Site Report: A Look at This Year’s Achievements and Future AI Trends - Company: LG AI Research (lgresearch.ai) - Announced: 2024-11-21 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=501 - Record: https://forck.live/items/4819-lg-ai-insight-2024-on-site-report-a-look-at-this-year-s-achievements-and - Subject: EXAONE LG AI Research held its annual LG AI Insight 2024 conference, recapping the past year's accomplishments and discussing future AI trends, including AI agents, LLMs, multimodal AI, bio/medical AI, and computing infrastructure. The event featured an AI Tech Seminar with academic and industry experts, a poster session showcasing 66 research posters, and special programs such as an Open Insight Office and a demo zone. Demos included ChatEXAONE, an AI agent for enterprise users, and the EXAONE model. ### Replicate — FLUX.1 Tools – Control and steerability for FLUX - Company: Replicate (replicate.com) - Announced: 2024-11-21 - Category: model-update - Coverage: not counted - Announcement: yes - Group: models - Source: https://replicate.com/blog/flux-tools - Record: https://forck.live/items/2507-flux-1-tools-control-and-steerability-for-flux - Subject: Platform Black Forest Labs released FLUX.1 Tools, a set of new models that add control and steerability to their FLUX text-to-image model. The lineup includes four features: Fill (inpainting/outpainting), Canny (edge detection), Depth (depth maps), and Redux (image variation adapter). Each feature is available for both FLUX.1 [dev] and FLUX.1 [pro], with Redux also available for FLUX.1 [schnell]. All models are now on Replicate. ### OpenAI — Building smarter maps with GPT-4o vision fine-tuning - Company: OpenAI (openai.com) - Announced: 2024-11-20T17:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/grab - Record: https://forck.live/items/670-building-smarter-maps-with-gpt-4o-vision-fine-tuning - Subject: GPT / ChatGPT / API The headline suggests using GPT-4o vision fine-tuning to build smarter maps. ### Vercel — Vercel acquires Grep to accelerate code search - Company: Vercel (vercel.com) - Announced: 2024-11-20T13:00:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://vercel.com/blog/vercel-acquires-grep - Record: https://forck.live/items/20922-vercel-acquires-grep-to-accelerate-code-search - Subject: v0 / AI SDK / AI Gateway Vercel acquired Grep, a tool for searching code across over 500,000 public git repositories. Grep's founder Dan Fox will join Vercel's AI team to continue developing the tool. ### Hugging Face — Introducing the Open Leaderboard for Japanese LLMs! - Company: Hugging Face (huggingface.co) - Announced: 2024-11-20 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/leaderboard-japanese - Record: https://forck.live/items/1797-introducing-the-open-leaderboard-for-japanese-llms - Subject: Platform Hugging Face and LLM-jp announce the Open Japanese LLM Leaderboard, a platform to evaluate and compare Japanese LLMs on over 20 NLP tasks. ### Hugging Face — Faster Text Generation with Self-Speculative Decoding - Company: Hugging Face (huggingface.co) - Announced: 2024-11-20 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/layerskip - Record: https://forck.live/items/1796-faster-text-generation-with-self-speculative-decoding - Subject: Platform Self-speculative decoding, a technique that uses early layers of a large language model for drafting tokens and later layers for verification, achieving faster text generation and memory savings. It provides implementation details using the transformers library and benchmarks comparing with traditional speculative decoding. ### Hugging Face — From Files to Chunks: Improving HF Storage Efficiency - Company: Hugging Face (huggingface.co) - Announced: 2024-11-20 - Category: infrastructure-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/from-files-to-chunks - Record: https://forck.live/items/1795-from-files-to-chunks-improving-hf-storage-efficiency - Subject: Platform Hugging Face announces a new storage approach using content-defined chunking (CDC) to improve storage efficiency and iteration speed. By storing files as chunks and transferring only modified chunks, the method reduces upload/download times and storage costs. Benchmarks show a 50% improvement in storage and transfer performance, and initial research on the Hub indicates potential savings of up to 100 TB for PyTorch model checkpoints. ### Hugging Face — Letting Large Models Debate: The First Multilingual LLM Debate Competition - Company: Hugging Face (huggingface.co) - Announced: 2024-11-20 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/debate - Record: https://forck.live/items/1794-letting-large-models-debate-the-first-multilingual-llm-debate-competition - Subject: Platform BAAI introduces FlagEval Debate, a multilingual debate platform for evaluating LLMs through direct debate competitions in English, Chinese, Arabic, and Korean, with dual evaluation metrics and developer customization. ### OpenAI — Rox goes “all in” on OpenAI - Company: OpenAI (openai.com) - Announced: 2024-11-19T07:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/rox - Record: https://forck.live/items/671-rox-goes-all-in-on-openai - Subject: GPT / ChatGPT / API Rox uses OpenAI's models to help sellers perform at a top 1% level. ### Suno — Introducing v4 - Company: Suno (suno.com) - Announced: 2024-11-19 - Category: model-update - Coverage: not counted - Announcement: yes - Group: models - Source: https://suno.com/blog/v4 - Record: https://forck.live/items/10337-introducing-v4 - Subject: Music / Studio Suno released v4, a major update to its music generation model, featuring cleaner audio, sharper lyrics, and more dynamic song structures. The update introduces a new lyrics assist model called ReMi and a Remaster feature to upgrade tracks from older models to v4 quality. v4 is available in beta to Pro and Premier subscribers on the web and iOS. ### LG AI Research — [AI Ethics Seminar EP.3] Introduction to Key AI Toolkits for Designing Trustworthy AI Experiences and Practical Application Guide - Company: LG AI Research (lgresearch.ai) - Announced: 2024-11-19 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=499 - Record: https://forck.live/items/4820-ai-ethics-seminar-ep-3-introduction-to-key-ai-toolkits-for-designing - Subject: EXAONE LG AI Research's AI Ethics Seminar discusses key AI toolkits (Microsoft HAX and Google PAIR) for designing trustworthy AI experiences and introduces LG's own AI Ethical Impact Assessment system. ### Hugging Face — Judge Arena: Benchmarking LLMs as Evaluators - Company: Hugging Face (huggingface.co) - Announced: 2024-11-19 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/arena-atla - Record: https://forck.live/items/1798-judge-arena-benchmarking-llms-as-evaluators - Subject: Platform Hugging Face and AtlaAI announce Judge Arena, a platform for benchmarking LLMs as evaluators. Users can compare models side-by-side by having them judge responses and voting on the best evaluation. The platform includes 18 models and will produce a public leaderboard. Early results show GPT-4 Turbo leading, but open-source models like Llama and Qwen are competitive. ### Vercel — AI SDK 4.0 - Company: Vercel (vercel.com) - Announced: 2024-11-18T13:00:00+00:00 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://vercel.com/blog/ai-sdk-4-0 - Record: https://forck.live/items/20923-ai-sdk-4-0 - Subject: v0 / AI SDK / AI Gateway Vercel released AI SDK 4.0, adding PDF support for Anthropic, Google Generative AI, and Google Vertex AI providers, computer use support for the latest Claude Sonnet 3.5 model, and an xAI Grok provider. ### Mistral AI — Pixtral Large - Company: Mistral AI (mistral.ai) - Announced: 2024-11-18 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://mistral.ai/news/pixtral-large - Record: https://forck.live/items/3337-pixtral-large - Subject: Mistral / Le Chat Mistral AI team publishes a research post titled 'Pixtral Large'. ### Mistral AI — Mistral has entered the chat - Company: Mistral AI (mistral.ai) - Announced: 2024-11-18 - Category: product-launch - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://mistral.ai/news/mistral-chat - Record: https://forck.live/items/3336-mistral-has-entered-the-chat - Subject: Mistral / Le Chat Mistral AI announces that Mistral has entered the chat on November 18, 2024.