# forck.live — 5–11 May 2025

> 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 2025-W19, 5–11 May 2025, 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/2025-W19
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### Sakana AI — Introducing Continuous Thought Machines

- Company: Sakana AI (sakana.ai)
- Announced: 2025-05-11T15:00:00+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://sakana.ai/ctm
- Record: https://forck.live/items/4718-introducing-continuous-thought-machines
- Subject: Sakana models

Sakana AI announced the Continuous Thought Machine (CTM), a new neural network model that uses neuron synchronization for reasoning, inspired by biological brains. The model shows interpretable behavior and improved performance on various tasks, and the code has been released.

### Hugging Face — LeRobot Community Datasets: The “ImageNet” of Robotics — When and How?

- Company: Hugging Face (huggingface.co)
- Announced: 2025-05-11
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/lerobot-datasets
- Record: https://forck.live/items/1701-lerobot-community-datasets-the-imagenet-of-robotics-when-and-how
- Subject: Platform

The importance of community-contributed datasets for advancing robot generalization, framing generalization as a data problem. It highlights the growing number of LeRobot datasets on the Hugging Face Hub, the challenges of data collection, and best practices for maximizing impact.

### Luma — Introducing Reframe

- Company: Luma (lumalabs.ai)
- Announced: 2025-05-09
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lumalabs.ai/news/introducing-reframe
- Record: https://forck.live/items/7986-introducing-reframe
- Subject: Ray / Dream Machine

Luma introduces Reframe, a feature for outpaint and resizing videos and images, generating new visual content at the edges while preserving the main subject. It is available on all paid plans.

### Amp — History in CLI

- Company: Amp (ampcode.com)
- Announced: 2025-05-09
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/cli-history
- Record: https://forck.live/items/7845-history-in-cli
- Subject: Amp

Amp has added the ability to navigate the local history of messages in its CLI using arrow keys, Ctrl+P/N, PageUp/Down, and Ctrl+C/Esc to cancel.

### Replit — Introducing Replit Auth: add secure login to your app

- Company: Replit (replit.com)
- Announced: 2025-05-08T16:35:28+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/auth
- Record: https://forck.live/items/17960-introducing-replit-auth-add-secure-login-to-your-app
- Subject: Replit Agent

Replit introduced Replit Auth, a zero-setup authentication system that developers can add to their apps by including "use Replit Auth" in their prompt. It provides user login, social sign-in, user management, and enterprise-grade security, leveraging existing Replit accounts and OpenID Connect. The feature is available now at no extra cost.

### Ant Ling — Ling: A MoE LLM Provided and Open-sourced by InclusionAI

- Company: Ant Ling (ant-ling.com)
- Announced: 2025-05-08
- Category: open-weight-release
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://developer.ant-ling.com/en/blogs/ling
- Record: https://forck.live/items/22347-ling-a-moe-llm-provided-and-open-sourced-by-inclusionai
- Subject: Ling / Ring / Ming

InclusionAI open-sourced Ling, a mixture-of-experts large language model, in two sizes: Ling-lite (16.8B total, 2.75B activated parameters) and Ling-plus (290B total, 28.8B activated parameters). The models are available on Hugging Face and ModelScope. A subsequent update released Ling-lite-1.5 with improved reasoning, and Ling-lite-0415 with improvements in code and math over the earlier Ling-lite-0220.

### Supabase — PostgreSQL Event Triggers without superuser access

- Company: Supabase (supabase.com)
- Announced: 2025-05-08
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/event-triggers-wo-superuser
- Record: https://forck.live/items/21359-postgresql-event-triggers-without-superuser-access
- Subject: Supabase

Supabase's supautils extension now allows regular users to create PostgreSQL Event Triggers without superuser access, using a privileged role and hooks to safely skip triggers for superusers and reserved roles. The feature is available for new projects on the Supabase platform and can be installed from the supautils repo.

### Genspark — Introducing Genspark AI Sheets - Welcome to The Full Agentic Data Auto-search or Analysis Experience!

- Company: Genspark (genspark.ai)
- Announced: 2025-05-08
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.genspark.ai/blog/genspark-ai-sheets
- Record: https://forck.live/items/18287-introducing-genspark-ai-sheets-welcome-to-the-full-agentic-data-auto-search-or
- Subject: Genspark

### Amp — No More BYOK

- Company: Amp (ampcode.com)
- Announced: 2025-05-08
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/no-more-byok
- Record: https://forck.live/items/7846-no-more-byok
- Subject: Amp

Amp removed Isolated Mode (bring-your-own-keys) and now supports customer-managed model provider connections that allow users to configure API keys and ChatGPT Plus/Pro subscriptions directly in settings.

### OpenAI — OpenAI Expands Leadership with Fidji Simo

- Company: OpenAI (openai.com)
- Announced: 2025-05-07T21:00:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/leadership-expansion-with-fidji-simo
- Record: https://forck.live/items/564-openai-expands-leadership-with-fidji-simo
- Subject: GPT / ChatGPT / API

OpenAI announced that Sam shared a message with the company.

### OpenAI — OpenAI’s response to the Department of Energy on AI infrastructure

- Company: OpenAI (openai.com)
- Announced: 2025-05-07T18:30:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/global-affairs/response-to-department-of-energy
- Record: https://forck.live/items/565-openai-s-response-to-the-department-of-energy-on-ai-infrastructure
- Subject: GPT / ChatGPT / API

OpenAI responds to the Department of Energy on AI infrastructure, arguing that infrastructure is destiny and explaining how the US can seize it.

### OpenAI — Introducing data residency in Asia

- Company: OpenAI (openai.com)
- Announced: 2025-05-07T18:00:00+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/introducing-data-residency-in-asia
- Record: https://forck.live/items/566-introducing-data-residency-in-asia
- Subject: GPT / ChatGPT / API

OpenAI introduces data residency in Asia, building on existing enterprise-grade data privacy, security, and compliance programs.

### Vercel — Introducing the Flags Explorer, first-party integrations, and updates to the Flags SDK

- Company: Vercel (vercel.com)
- Announced: 2025-05-07T13:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/introducing-the-flags-explorer-first-party-integrations-and-updates
- Record: https://forck.live/items/20878-introducing-the-flags-explorer-first-party-integrations-and-updates-to-the
- Subject: v0 / AI SDK / AI Gateway

Vercel made the Flags Explorer generally available for all plans, allowing users to manage and override feature flags directly from the Vercel Toolbar. The company also introduced a new Experimentation category on the Vercel Marketplace with Statsig as the launch partner, and updated the Flags SDK with a new adapters architecture for loading flags from providers.

### OpenAI — The San Antonio Spurs use ChatGPT to scale impact on and off the court

- Company: OpenAI (openai.com)
- Announced: 2025-05-07T09:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/san-antonio-spurs
- Record: https://forck.live/items/567-the-san-antonio-spurs-use-chatgpt-to-scale-impact-on-and-off-the-court
- Subject: GPT / ChatGPT / API

OpenAI published a case study about the San Antonio Spurs using custom GPTs built on ChatGPT.

### OpenAI — Lowe’s puts project expertise into every hand

- Company: OpenAI (openai.com)
- Announced: 2025-05-07T07:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/lowes
- Record: https://forck.live/items/568-lowe-s-puts-project-expertise-into-every-hand
- Subject: GPT / ChatGPT / API

Lowe’s has partnered with OpenAI to build two AI-powered tools, Mylow and Mylow Companion, designed to help customers and store associates plan, navigate, and complete home improvement projects.

### OpenAI — Introducing OpenAI for Countries

- Company: OpenAI (openai.com)
- Announced: 2025-05-07T03:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/global-affairs/openai-for-countries
- Record: https://forck.live/items/569-introducing-openai-for-countries
- Subject: GPT / ChatGPT / API

OpenAI announces a new initiative to support countries that want to build on democratic AI rails.

### Ant Ling — Ming-Lite-Uni: Advancements in Unified Architecture for Natural Multimodal Interaction

- Company: Ant Ling (ant-ling.com)
- Announced: 2025-05-07
- Category: open-weight-release
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://developer.ant-ling.com/en/blogs/ming-lite-uni
- Record: https://forck.live/items/22351-ming-lite-uni-advancements-in-unified-architecture-for-natural-multimodal
- Subject: Ling / Ring / Ming

Ant Ling released Ming-Lite-Uni, an open-source multimodal framework that integrates a unified visual generator with a native multimodal autoregressive model for vision and language tasks. The framework supports text-to-image generation and instruction-based image editing, achieving an average understanding score of 69.7 on OpenCompass and an image generation score of 0.62 on GenEval. It is currently in alpha phase and available with full model weights and test code.

### Amp — File Mentions in CLI

- Company: Amp (ampcode.com)
- Announced: 2025-05-07
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/cli-file-mentions
- Record: https://forck.live/items/7848-file-mentions-in-cli
- Subject: Amp

Amp added file-mention support to its CLI interactive mode, enabling users to fuzzy-search files by typing @ followed by a pattern and navigating results with Tab or Shift-Tab.

### Amp — AGENT.md

- Company: Amp (ampcode.com)
- Announced: 2025-05-07
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/AGENT.md
- Record: https://forck.live/items/7847-agent-md
- Subject: Amp

Amp now reads the AGENT.md file at the project root for guidance on structure, build steps, and conventions, and can generate this file by scanning existing agent configuration files.

### Mistral AI — Medium is the new large.

- Company: Mistral AI (mistral.ai)
- Announced: 2025-05-07
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://mistral.ai/news/mistral-medium-3
- Record: https://forck.live/items/3326-medium-is-the-new-large
- Subject: Mistral / Le Chat

Mistral AI announces that medium is the new large.

### Mistral AI — Introducing Le Chat Enterprise

- Company: Mistral AI (mistral.ai)
- Announced: 2025-05-07
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://mistral.ai/news/le-chat-enterprise
- Record: https://forck.live/items/3325-introducing-le-chat-enterprise
- Subject: Mistral / Le Chat

Mistral AI announced the introduction of Le Chat Enterprise.

### Replicate — Ideogram 3.0 on Replicate

- Company: Replicate (replicate.com)
- Announced: 2025-05-07
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://replicate.com/blog/ideogram-v3
- Record: https://forck.live/items/2496-ideogram-3-0-on-replicate
- Subject: Platform

Ideogram 3.0 is a major update to the text-to-image model, with three varieties: Turbo, Balanced, and Quality. It offers improvements in realism, style control, and layout generation, and is available on Replicate.

### Harvey — The Brief: May 2025

- Company: Harvey (harvey.ai)
- Announced: 2025-05-06T15:28:40.866000+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/the-brief-may-2025
- Record: https://forck.live/items/17711-the-brief-may-2025
- Subject: Harvey

Harvey rolled out GPT-4.1 as the primary model for Assist mode and Vault, and enhanced Draft mode with a reasoning model, improving response quality, output length, and knowledge cutoff. The company also introduced Vault Knowledge Bases for creating internal knowledge bases, along with various UI and workflow improvements.

### Vercel — Join the Vercel AI Accelerator

- Company: Vercel (vercel.com)
- Announced: 2025-05-06T13:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://vercel.com/blog/join-the-vercel-ai-accelerator
- Record: https://forck.live/items/20879-join-the-vercel-ai-accelerator
- Subject: v0 / AI SDK / AI Gateway

Vercel announced the return of its AI Accelerator, a six-week program for 40 teams offering over $4 million in credits from Vercel, v0, AWS, and other AI platforms. Applications are open until May 17, and the program runs from June 9 to July 18, ending with a demo day where three winners receive over $100K in credits and the first-place winner also receives an investment from Vercel Ventures.

### OpenAI — Introducing AI stories: daily benefits shine a light on bigger opportunities

- Company: OpenAI (openai.com)
- Announced: 2025-05-06T10:30:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/global-affairs/ai-stories-daily-benefits-bigger-opportunities
- Record: https://forck.live/items/570-introducing-ai-stories-daily-benefits-shine-a-light-on-bigger-opportunities
- Subject: GPT / ChatGPT / API

The text is a general vision statement about the potential of AI to solve big problems, not a specific product or model announcement.

### Cartesia — Introducing Professional Voice Cloning

- Company: Cartesia (cartesia.ai)
- Announced: 2025-05-06
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.cartesia.ai/blog/pro-voice-cloning
- Record: https://forck.live/items/8019-introducing-professional-voice-cloning
- Subject: Sonic / Ink

Cartesia introduced Professional Voice Clones (PVCs) built on the Sonic text-to-speech model, available on the Startup plan and above. PVCs are created by fine-tuning Sonic on voice data, enabling replicas of tone, cadence, style, and environment. Training is free, and generating PVC speech costs 1 credit per character.

### Replicate — Run MiniMax Speech-02 models with an API

- Company: Replicate (replicate.com)
- Announced: 2025-05-06
- Category: api-release
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replicate.com/blog/minimax-text-to-speech
- Record: https://forck.live/items/2497-run-minimax-speech-02-models-with-an-api
- Subject: Platform

Replicate announces API availability for MiniMax Speech-02 text-to-speech models (HD and Turbo), with voice cloning, emotion control, and support for 30+ languages. Pricing is $30 per million characters for Turbo, $50 per million for HD, and $3 per voice clone.

### OpenAI — AI helps John Deere transform agriculture

- Company: OpenAI (openai.com)
- Announced: 2025-05-06
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/john-deere-justin-rose
- Record: https://forck.live/items/571-ai-helps-john-deere-transform-agriculture
- Subject: GPT / ChatGPT / API

A discussion with John Deere's Justin Rose about using AI to transform agriculture and scale innovation for farmers.

### OpenAI — Evolving OpenAI’s structure

- Company: OpenAI (openai.com)
- Announced: 2025-05-05T11:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/evolving-our-structure
- Record: https://forck.live/items/572-evolving-openai-s-structure
- Subject: GPT / ChatGPT / API

OpenAI's board announced the transition of its for-profit entity to a Public Benefit Corporation to reinforce its mission-driven structure under nonprofit oversight.

### OpenAI — Lowe’s leverages AI to power home improvement retail

- Company: OpenAI (openai.com)
- Announced: 2025-05-05T05:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/lowes-chandhu-nair
- Record: https://forck.live/items/573-lowe-s-leverages-ai-to-power-home-improvement-retail
- Subject: GPT / ChatGPT / API

A conversation with Chandhu Nair, Senior Vice President of Data, AI, and Innovation at Lowe's, about leveraging AI in home improvement retail.

### Ant Ling — Ming-Lite-Omni-Preview: A MoE Model Designed to Perceive a Wide Range of Modalities

- Company: Ant Ling (ant-ling.com)
- Announced: 2025-05-05
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://developer.ant-ling.com/en/blogs/ming-lite-omni-preview
- Record: https://forck.live/items/22350-ming-lite-omni-preview-a-moe-model-designed-to-perceive-a-wide-range-of
- Subject: Ling / Ring / Ming

Ant Ling released Ming-Lite-Omni-Preview, a Mixture-of-Experts model built on Ling-Lite that processes text, images, audio, and video while generating text and natural speech in a streaming manner. The model incorporates modality-specific routers and supports video understanding with KV-Cache dynamic compression, natural speech generation with dialect support and one-shot voice cloning, and a chain-of-thought reasoning mode. It is available for download from Hugging Face and ModelScope.
