# forck.live — 1–7 April 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-W14, 1–7 April 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-W14 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Replit — Advanced port configuration - Company: Replit (replit.com) - Announced: 2024-04-05T15:00:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replit.com/blog/ports - Record: https://forck.live/items/17999-advanced-port-configuration - Subject: Replit Agent Replit introduced explicit port mappings stored in the .replit config file, replacing the previous automatic forwarding of the first open port. This change makes port assignments predictable and persistent, enabling users to run multiple services in the same Repl and ensuring forked projects work identically. ### OpenAI — Klarna's AI assistant does the work of 700 full-time agents - Company: OpenAI (openai.com) - Announced: 2024-04-05 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/klarna - Record: https://forck.live/items/787-klarna-s-ai-assistant-does-the-work-of-700-full-time-agents - Subject: GPT / ChatGPT / API Klarna is using AI to transform personal shopping, customer service, and employee productivity. ### Vercel — Improved infrastructure pricing - Company: Vercel (vercel.com) - Announced: 2024-04-04T13:00:00+00:00 - Category: pricing-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://vercel.com/blog/improved-infrastructure-pricing - Record: https://forck.live/items/20996-improved-infrastructure-pricing - Subject: v0 / AI SDK / AI Gateway Vercel announced updated infrastructure pricing that splits previously bundled bandwidth and functions metrics into separate, granular components. Bandwidth pricing drops from $0.40/GB to $0.15/GB for Fast Data Transfer and $0.06/GB for Fast Origin Transfer; functions pricing changes from $0.40/GB-hr to $0.60 per million invocations plus $0.18/GB-hr for duration. The Hobby tier remains free, and the Pro plan's $20/month base now includes 1 TB Fast Data Transfer, 100 GB Fast Origin Transfer, and 10 million Edge Requests before on-demand charges apply. ### Upstage — Solar LLM rises in the east via AWS SageMaker Jumpstart - Company: Upstage (upstage.ai) - Announced: 2024-04-04 - Category: availability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.upstage.ai/blog/en/aws-sagemaker-jumpstart-solar-llm - Record: https://forck.live/items/18485-solar-llm-rises-in-the-east-via-aws-sagemaker-jumpstart - Subject: Solar Pro / Solar Mini Upstage announced that its Solar LLM is now available on Amazon SageMaker JumpStart, allowing users to build customizable generative AI services without creating their own models. Solar, which topped the Open LLM Leaderboard of Hugging Face in December 2023, uses a depth up-scaling method and is based on a 32-layer Llama 2 architecture initialized with Mistral 7B weights. The post also describes Upstage's document processing components including Layout Analysis, RAG, and Groundedness Check to reduce hallucinations. ### Hugging Face — Text2SQL using Hugging Face Dataset Viewer API and Motherduck DuckDB-NSQL-7B - Company: Hugging Face (huggingface.co) - Announced: 2024-04-04 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://huggingface.co/blog/duckdb-nsql-7b - Record: https://forck.live/items/1915-text2sql-using-hugging-face-dataset-viewer-api-and-motherduck-duckdb-nsql-7b - Subject: Platform MotherDuck and Numbers Station announced DuckDB-NSQL-7B, a large language model fine-tuned from Llama-2-7b for generating DuckDB SQL queries. ### Hugging Face — Hugging Face partners with Wiz Research to Improve AI Security - Company: Hugging Face (huggingface.co) - Announced: 2024-04-04 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/hugging-face-wiz-security-blog - Record: https://forck.live/items/1914-hugging-face-partners-with-wiz-research-to-improve-ai-security - Subject: Platform Hugging Face announces a partnership with Wiz Research to improve security across its platform and the AI/ML ecosystem, including vulnerability management and cloud security posture management. ### OpenAI — Introducing improvements to the fine-tuning API and expanding our custom models program - Company: OpenAI (openai.com) - Announced: 2024-04-04 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/introducing-improvements-to-the-fine-tuning-api-and-expanding-our-custom-models-program - Record: https://forck.live/items/788-introducing-improvements-to-the-fine-tuning-api-and-expanding-our-custom - Subject: GPT / ChatGPT / API OpenAI is adding new features to the fine-tuning API and expanding its custom models program. ### Hugging Face — Blazing Fast SetFit Inference with 🤗 Optimum Intel on Xeon - Company: Hugging Face (huggingface.co) - Announced: 2024-04-03 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/setfit-optimum-intel - Record: https://forck.live/items/1916-blazing-fast-setfit-inference-with-optimum-intel-on-xeon - Subject: Platform Hugging Face and Intel announce that SetFit inference can be accelerated by 7.8x on Intel CPUs using Optimum Intel, an open-source library that includes post-training static quantization to convert models to INT8 precision, enabling production-grade deployment on Intel Xeon CPUs. ### Replit — Get early access to Replit Teams - Company: Replit (replit.com) - Announced: 2024-04-02T21:32:00+00:00 - Category: availability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replit.com/blog/teams-beta - Record: https://forck.live/items/18000-get-early-access-to-replit-teams - Subject: Replit Agent Replit announced early access to Replit Teams, a new product for collaborative software development with AI. The beta includes features such as AI-powered code completion and chat, ultra-fast workspaces with 8vCPUs and 16GB RAM, shared deployments, living repos, role-based access control, and centralized billing. Customers can join a waitlist for early access. ### Replit — Building LLMs for Code Repair - Company: Replit (replit.com) - Announced: 2024-04-02T21:26:33+00:00 - Category: research-paper - Coverage: not counted - Announcement: no - Group: routine - Source: https://replit.com/blog/code-repair - Record: https://forck.live/items/17998-building-llms-for-code-repair - Subject: Replit Agent Replit describes its methodology for training a code repair model using LSP diagnostics and synthetic diffs, aiming to create a Replit-native AI model that can fix code errors directly within the development environment. ### Alibaba — Qwen1.5-32B: Fitting the Capstone of the Qwen1.5 Language Model Series - Company: Alibaba (alibaba.com) - Announced: 2024-04-02T05:33:00+00:00 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://qwenlm.github.io/blog/qwen1.5-32b/ - Record: https://forck.live/items/2317-qwen1-5-32b-fitting-the-capstone-of-the-qwen1-5-language-model-series - Subject: Qwen Alibaba announces a new model with approximately 30 billion parameters, aiming to balance performance, efficiency, and memory footprint. ### Hugging Face — Bringing serverless GPU inference to Hugging Face users - Company: Hugging Face (huggingface.co) - Announced: 2024-04-02 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/cloudflare-workers-ai - Record: https://forck.live/items/1917-bringing-serverless-gpu-inference-to-hugging-face-users - Subject: Platform Hugging Face announces a new integration with Cloudflare Workers AI, allowing users to deploy open models as a serverless API with pay-per-request pricing, powered by GPUs in Cloudflare edge data centers. ### OpenAI — Customizing models for legal professionals - Company: OpenAI (openai.com) - Announced: 2024-04-02 - Category: partnership-acquisition - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/harvey - Record: https://forck.live/items/789-customizing-models-for-legal-professionals - Subject: GPT / ChatGPT / API Harvey and OpenAI announced a partnership to build a custom-trained model for legal professionals. ### Supabase — Announcing Data Preservation Service - Company: Supabase (supabase.com) - Announced: 2024-04-01 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://supabase.com/blog/pg-paper-dump - Record: https://forck.live/items/21442-announcing-data-preservation-service - Subject: Supabase We live in a world where data is as critical as air and water, powering everything from global enterprises to personal projects. At Supabase, we create products that are not just cutting-edge but also secure, reliable, and now, timeless. ### LG AI Research — Why AI Ethicists Constantly Challenge - Company: LG AI Research (lgresearch.ai) - Announced: 2024-04-01 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/news/view?seq=420 - Record: https://forck.live/items/11030-why-ai-ethicists-constantly-challenge - Subject: EXAONE LG AI Research's AI ethicists Soyoung An and Myoungshin Kim discuss current AI ethics topics including safety, fairness, and bias in AI systems. They highlight the EU AI Act and the establishment of AI safety institutes in the US, UK, and Japan as key developments. The ethicists also note the challenge of balancing universal human values with country-specific historical and cultural contexts when implementing fair AI. ### LG AI Research — Why AI Ethicists Constantly Challenge - Company: LG AI Research (lgresearch.ai) - Announced: 2024-04-01 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=420 - Record: https://forck.live/items/4859-why-ai-ethicists-constantly-challenge - Subject: EXAONE LG AI Research interviews its AI ethicists Soyoung An and Myoungshin Kim about their work, the importance of AI ethics, current hot topics such as safety and fairness, and the challenges of balancing transparency with misuse risks. ### OpenAI — Reducing health insurance costs and improving care - Company: OpenAI (openai.com) - Announced: 2024-04-01 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/oscar - Record: https://forck.live/items/791-reducing-health-insurance-costs-and-improving-care - Subject: GPT / ChatGPT / API Oscar is using AI to reduce health insurance costs and improve patient care. ### OpenAI — Start using ChatGPT instantly - Company: OpenAI (openai.com) - Announced: 2024-04-01 - Category: availability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/start-using-chatgpt-instantly - Record: https://forck.live/items/790-start-using-chatgpt-instantly - Subject: GPT / ChatGPT / API OpenAI announced that users can now experience the benefits of AI without needing to sign up, making access easier.