# forck.live — 30 September – 6 October 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-W40, 30 September – 6 October 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-W40 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Hugging Face — Improving Parquet Dedupe on Hugging Face Hub - Company: Hugging Face (huggingface.co) - Announced: 2024-10-05 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/improve_parquet_dedupe - Record: https://forck.live/items/1820-improving-parquet-dedupe-on-hugging-face-hub - Subject: Platform The Xet team at Hugging Face is researching and experimenting with techniques to improve Parquet file deduplication on the Hugging Face Hub, including content-defined row groups and relative offsets. ### Vercel — Navigating Web3 dynamism: Ledger's solution to traffic spike stability with Vercel - Company: Vercel (vercel.com) - Announced: 2024-10-04T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/ledgers-solution-to-traffic-spike-stability-with-vercel - Record: https://forck.live/items/20942-navigating-web3-dynamism-ledger-s-solution-to-traffic-spike-stability-with - Subject: v0 / AI SDK / AI Gateway Ledger migrated its ecommerce and web applications from a Shopify-based stack to Next.js and Vercel, achieving a 67% reduction in load time (from 2.6 to under 0.93 seconds), a 100x increase in release frequency, and a 13% improvement in bounce rates. The new infrastructure handles 6–7 million daily requests through a unified API platform. ### Hugging Face — Introducing the Open FinLLM Leaderboard - Company: Hugging Face (huggingface.co) - Announced: 2024-10-04 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/leaderboard-finbench - Record: https://forck.live/items/1821-introducing-the-open-finllm-leaderboard - Subject: Platform Hugging Face introduces the Open FinLLM Leaderboard, a specialized evaluation framework for financial language models (LLMs) that goes beyond general NLP benchmarks. It focuses on finance-specific tasks such as information extraction, sentiment analysis, credit risk scoring, and stock movement forecasting, using zero-shot evaluation and diverse metrics. ### Vercel — Serverless servers: Efficient serverless Node.js with in-function concurrency - Company: Vercel (vercel.com) - Announced: 2024-10-03T13:00:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://vercel.com/blog/serverless-servers-node-js-with-in-function-concurrency - Record: https://forck.live/items/20943-serverless-servers-efficient-serverless-node-js-with-in-function-concurrency - Subject: v0 / AI SDK / AI Gateway Vercel announced a new version of Vercel Functions with in-function concurrency, now in public beta. The feature allows a single function instance to handle multiple calls simultaneously, leveraging idle time to reduce compute usage and costs by 20-50% without latency impact, based on testing with private beta customers. The runtime has been rewritten in Rust to support this architecture. ### OpenAI — New Credit Facility Enhances Financial Flexibility - Company: OpenAI (openai.com) - Announced: 2024-10-03T07:00:00+00:00 - Category: funding-company-news - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/new-credit-facility-enhances-financial-flexibility - Record: https://forck.live/items/684-new-credit-facility-enhances-financial-flexibility - Subject: GPT / ChatGPT / API OpenAI secured $6.6 billion in new funding from investors and established a $4 billion credit facility with major banks. ### Replicate — FLUX1.1 [pro] is here - Company: Replicate (replicate.com) - Announced: 2024-10-03 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://replicate.com/blog/flux-1-1-pro-is-here - Record: https://forck.live/items/2512-flux1-1-pro-is-here - Subject: Platform Replicate announces FLUX1.1 [pro], a new faster image generation model from Black Forest Labs, priced at $0.04 per image, and also updates FLUX.1 [pro] to be twice as fast, priced at $0.055 per image. ### Hugging Face — A Short Summary of Chinese AI Global Expansion - Company: Hugging Face (huggingface.co) - Announced: 2024-10-03 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/chinese-ai-expansion - Record: https://forck.live/items/1822-a-short-summary-of-chinese-ai-global-expansion - Subject: Platform Chinese AI companies expanding globally, discussing major players like Huawei, Tencent, Alibaba, ByteDance, and startups, as well as the competitive domestic market driving overseas expansion. ### Lovable — Better Version Management and Speed Enhancements - Company: Lovable (lovable.dev) - Announced: 2024-10-02T14:03:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://lovable.dev/blog/2025-01-13-better-version-management-and-speed-enhancements - Record: https://forck.live/items/12205-better-version-management-and-speed-enhancements - Subject: Lovable / AI app builder Lovable added descriptive labels and previews to each version, making it easier to locate and revert to past edits, and now provides quick access to the last 8 edits. Large projects with many edits are now up to 12 times faster because messages are rendered on demand as users scroll rather than all at once. The company also lowered the threshold for using RAG, reducing errors by 25% in smaller projects, and made GPT Engineer less eager to generate code immediately, allowing more exploratory conversations before code changes. ### Vercel — Vercel WAF upgrade brings persistent actions, rate limiting, and API control - Company: Vercel (vercel.com) - Announced: 2024-10-02T13:00:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://vercel.com/blog/vercel-waf-upgrade-brings-persistent-actions-rate-limiting-and-api-control - Record: https://forck.live/items/20944-vercel-waf-upgrade-brings-persistent-actions-rate-limiting-and-api-control - Subject: v0 / AI SDK / AI Gateway Vercel announced enhancements to its Web Application Firewall (WAF), including persistent actions that block repeat offenders for a customizable duration, rate limiting for specific endpoints to prevent API abuse, and a Firewall API for programmatic control of WAF settings. The Challenge page feature, previously exclusive to the WAF, is now extended to the broader Vercel Firewall. The WAF propagates rules globally in 300 milliseconds. ### OpenAI — New funding to scale the benefits of AI - Company: OpenAI (openai.com) - Announced: 2024-10-02T10:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/scale-the-benefits-of-ai - Record: https://forck.live/items/685-new-funding-to-scale-the-benefits-of-ai - Subject: GPT / ChatGPT / API OpenAI states it is making progress on its mission to ensure that artificial general intelligence benefits all of humanity. ### OpenAI — Introducing the Realtime API - Company: OpenAI (openai.com) - Announced: 2024-10-01T10:05:00+00:00 - Category: api-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/introducing-the-realtime-api - Record: https://forck.live/items/686-introducing-the-realtime-api - Subject: GPT / ChatGPT / API OpenAI announced the Realtime API, enabling developers to integrate fast speech-to-speech experiences into their applications. ### OpenAI — Introducing vision to the fine-tuning API - Company: OpenAI (openai.com) - Announced: 2024-10-01T10:04:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/introducing-vision-to-the-fine-tuning-api - Record: https://forck.live/items/687-introducing-vision-to-the-fine-tuning-api - Subject: GPT / ChatGPT / API OpenAI announced that developers can now fine-tune GPT-4o using both images and text to enhance its vision capabilities. ### OpenAI — Prompt Caching in the API - Company: OpenAI (openai.com) - Announced: 2024-10-01T10:03:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/api-prompt-caching - Record: https://forck.live/items/688-prompt-caching-in-the-api - Subject: GPT / ChatGPT / API OpenAI introduces automatic discounts for repeated input tokens in its API. ### OpenAI — Model Distillation in the API - Company: OpenAI (openai.com) - Announced: 2024-10-01T10:02:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/api-model-distillation - Record: https://forck.live/items/689-model-distillation-in-the-api - Subject: GPT / ChatGPT / API OpenAI announces model distillation in the API, allowing users to fine-tune a cost-efficient model using outputs from a large frontier model, all within the OpenAI platform. ### OpenAI — Creating agent and human collaboration with GPT 4o - Company: OpenAI (openai.com) - Announced: 2024-10-01T09:59:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/altera - Record: https://forck.live/items/690-creating-agent-and-human-collaboration-with-gpt-4o - Subject: GPT / ChatGPT / API Altera uses GPT-4o to develop a new form of human collaboration. ### Microsoft AI — An AI companion for everyone - Company: Microsoft AI (microsoft.ai) - Announced: 2024-10-01 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://microsoft.ai/news/an-ai-companion-for-everyone - Record: https://forck.live/items/8440-an-ai-companion-for-everyone - Subject: MAI Microsoft AI's CEO Mustafa Suleyman outlined a vision for Copilot as an AI companion that will adapt to users, safeguard privacy, and act on their behalf with permission, aiming to enrich human lives and strengthen bonds. ### Hugging Face — 🇨🇿 BenCzechMark - Can your LLM Understand Czech? - Company: Hugging Face (huggingface.co) - Announced: 2024-10-01 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/benczechmark - Record: https://forck.live/items/1823-benczechmark-can-your-llm-understand-czech - Subject: Platform Hugging Face announces the BenCzechMark, a comprehensive evaluation suite and leaderboard for testing LLMs on Czech language understanding, reasoning, and knowledge tasks. ### OpenAI — STORM-0817: Iran-linked malware and scraping activity - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-storm-0817 - Record: https://forck.live/items/701-storm-0817-iran-linked-malware-and-scraping-activity - Subject: GPT / ChatGPT / API OpenAI banned accounts linked to the Iran-based group STORM-0817 that were using AI for malicious purposes, including debugging Android malware, scraping social media platforms, and translating tools. ### OpenAI — SweetSpecter: China-linked cyber activity - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-sweetspecter - Record: https://forck.live/items/700-sweetspecter-china-linked-cyber-activity - Subject: GPT / ChatGPT / API OpenAI banned accounts associated with a suspected China-based adversary using AI for malicious activities. ### OpenAI — CyberAv3ngers: Iran-linked cyber research activity - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-cyberav3ngers - Record: https://forck.live/items/699-cyberav3ngers-iran-linked-cyber-research-activity - Subject: GPT / ChatGPT / API OpenAI banned accounts linked to CyberAv3ngers that were using AI to research industrial control systems, default credentials, and targets. ### OpenAI — Hoax: Fake Russian “troll” error message - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-hoax-russian-troll - Record: https://forck.live/items/698-hoax-fake-russian-troll-error-message - Subject: GPT / ChatGPT / API OpenAI banned an account that likely originated in the US and used AI to create a fake ChatGPT error message claiming to detect 'Russian troll' activity. ### OpenAI — Operation “Stop News”: Russia-origin influence activity - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-stop-news-2024 - Record: https://forck.live/items/697-operation-stop-news-russia-origin-influence-activity - Subject: GPT / ChatGPT / API OpenAI has banned a Russia-origin influence activity called 'Stop News', which used its AI to create multilingual content targeting Ukraine and the West. ### OpenAI — Operation “A2Z”: Multilingual influence activity - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-a2z - Record: https://forck.live/items/696-operation-a2z-multilingual-influence-activity - Subject: GPT / ChatGPT / API OpenAI banned accounts associated with the 'A2Z' operation that used AI to generate multilingual influence content related to elections, Ukraine, and politics across various platforms. ### OpenAI — Operation “STORM-2035”: Iran-origin influence activity - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-storm-2035-2024 - Record: https://forck.live/items/695-operation-storm-2035-iran-origin-influence-activity - Subject: GPT / ChatGPT / API OpenAI banned accounts linked to Iran-based influence operation STORM-2035 that used AI to generate election-related content targeting the US and UK and published it across sites. ### OpenAI — Bet Bot: Gambling spam network - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-bet-bot - Record: https://forck.live/items/694-bet-bot-gambling-spam-network - Subject: GPT / ChatGPT / API OpenAI banned accounts that used its models via an Israel-based startup to create conversations and post gambling-site links on X. ### OpenAI — Rwandan election content: Political commenting network - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-rwandan-election-content - Record: https://forck.live/items/693-rwandan-election-content-political-commenting-network - Subject: GPT / ChatGPT / API OpenAI banned accounts originating from Rwanda that were using AI to generate partisan comments in preparation for the country's elections. ### OpenAI — Tort Report: Abusive reporting activity - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-tort-report - Record: https://forck.live/items/692-tort-report-abusive-reporting-activity - Subject: GPT / ChatGPT / API OpenAI banned accounts that used AI to generate abusive reports and complaints against Vietnamese public figures and platforms. ### OpenAI — Corrupt Comment: Anti-corruption foundation criticism - Company: OpenAI (openai.com) - Announced: 2024-10-01 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/disrupting-malicious-uses-of-ai-corrupt-comment - Record: https://forck.live/items/691-corrupt-comment-anti-corruption-foundation-criticism - Subject: GPT / ChatGPT / API OpenAI banned accounts that used AI to generate comments criticizing a Russian anti-corruption foundation and related figures. ### Supabase — Supabase Launch Week 12 Hackathon Winners - Company: Supabase (supabase.com) - Announced: 2024-09-30 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://supabase.com/blog/lw12-hackathon-winners - Record: https://forck.live/items/21396-supabase-launch-week-12-hackathon-winners - Subject: Supabase Supabase announced the winners of its Launch Week 12 hackathon, with Whisker Jam winning best overall project. Other category winners include SupExplain (Best use of AI), Minecraft Server (Most fun / best easter egg), supagen (Most technically impressive), and Today I Learned (Most visually pleasing). ### LG AI Research — ‘LG Aimers’ has continued to grow, surpassing 12,000 participants - Company: LG AI Research (lgresearch.ai) - Announced: 2024-09-30 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=489 - Record: https://forck.live/items/4825-lg-aimers-has-continued-to-grow-surpassing-12-000-participants - Subject: EXAONE LG AI Research reports that its LG Aimers youth AI talent development program has surpassed 12,000 participants with the conclusion of the 5th hackathon, which involved a 24-hour competition using digital twin technology and real factory data, offering job screening exemptions and training subsidies to winners. ### Hugging Face — Converting Vertex-Colored Meshes to Textured Meshes - Company: Hugging Face (huggingface.co) - Announced: 2024-09-30 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/vertex-colored-to-textured-mesh - Record: https://forck.live/items/1824-converting-vertex-colored-meshes-to-textured-meshes - Subject: Platform To convert vertex-colored 3D meshes to UV-mapped, textured meshes using the InstantTexture library, with both a short version using a library call and a long version with manual implementation steps.