# forck.live — 14–20 July 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-W29, 14–20 July 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-W29
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### Supabase — Top 10 Launches of Launch Week 15

- Company: Supabase (supabase.com)
- Announced: 2025-07-18T22:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://supabase.com/blog/launch-week-15-top-10
- Record: https://forck.live/items/21341-top-10-launches-of-launch-week-15
- Subject: Supabase

Supabase announced ten product updates during Launch Week 15, including new API keys with asymmetric JWT support, Analytics Buckets with Apache Iceberg in private alpha, OpenTelemetry support, a Figma Make partnership, increased storage upload limits up to 500 GB and reduced cached egress pricing to $0.03/GB, Edge Functions with Deno 2.1 and up to 97% faster boot times, Branching 2.0 without GitHub dependency, a Platform Kit UI, a Stripe-to-Postgres sync engine as an npm package, and an Algolia connector.

### Supabase — Supabase Launch Week 15 Hackathon

- Company: Supabase (supabase.com)
- Announced: 2025-07-18T18:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://supabase.com/blog/lw15-hackathon
- Record: https://forck.live/items/21342-supabase-launch-week-15-hackathon
- Subject: Supabase

Supabase announced a hackathon for Launch Week 15, running from July 18 to July 27, 2025, with prizes in five categories including best overall project and best use of AI. Participants retain all rights to their submissions, which must be open source and use Supabase.

### Supabase — Storage: 10x Larger Uploads, 3x Cheaper Cached Egress, and 2x Egress Quota

- Company: Supabase (supabase.com)
- Announced: 2025-07-18T17:00:00+00:00
- Category: pricing-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/storage-500gb-uploads-cheaper-egress-pricing
- Record: https://forck.live/items/21343-storage-10x-larger-uploads-3x-cheaper-cached-egress-and-2x-egress-quota
- Subject: Supabase

Supabase increased the maximum file upload size from 50 GB to 500 GB for all paid plans, reduced cached egress pricing from $0.09/GB to $0.03/GB, and doubled the bundled egress quota for paid plans by splitting it into 250 GB cached and 250 GB uncached. The 500 GB upload limit becomes available next week; the new egress pricing and quotas roll out gradually over the next few weeks.

### Supabase — Persistent Storage and 97% Faster Cold Starts for Edge Functions

- Company: Supabase (supabase.com)
- Announced: 2025-07-18
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/persistent-storage-for-faster-edge-functions
- Record: https://forck.live/items/21344-persistent-storage-and-97-faster-cold-starts-for-edge-functions
- Subject: Supabase

Supabase introduced persistent storage for Edge Functions, allowing S3-compatible buckets to be mounted as a directory, and reduced cold start times by up to 97% by moving initial script evaluation to a dedicated blocking pool. The changes are deployed in all regions.

### Upstage — Document Intelligence now available in the new AWS Marketplace AI Agents and Tools category

- Company: Upstage (upstage.ai)
- Announced: 2025-07-18
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.upstage.ai/news/aws-agent-docai
- Record: https://forck.live/items/18430-document-intelligence-now-available-in-the-new-aws-marketplace-ai-agents-and
- Subject: Solar Pro / Solar Mini

Upstage made its Document Intelligence solution available in the new AI Agents and Tools category of AWS Marketplace. Customers can discover, buy, and deploy the intelligent document processing APIs using their AWS accounts. The solution converts unstructured documents into structured formats like HTML or Markdown for LLMs, supporting use cases such as contract ingestion, research-paper summarization, and invoice processing.

### Hugging Face — Arc Virtual Cell Challenge: A Primer

- Company: Hugging Face (huggingface.co)
- Announced: 2025-07-18
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/virtual-cell-challenge
- Record: https://forck.live/items/1653-arc-virtual-cell-challenge-a-primer
- Subject: Platform

This blog post serves as a primer for the Arc Virtual Cell Challenge, explaining the goal of training a model to predict the effect of gene silencing using CRISPR, describing the dataset, the modeling challenge, and the STATE baseline model released by Arc Institute.

### OpenAI — A $50 million fund to build with communities

- Company: OpenAI (openai.com)
- Announced: 2025-07-18
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/50-million-fund-to-build-with-communities
- Record: https://forck.live/items/524-a-50-million-fund-to-build-with-communities
- Subject: GPT / ChatGPT / API

OpenAI is launching an initial $50 million fund that supports nonprofit and community organizations, informed by the independent OpenAI Nonprofit Commission report.

### Harvey — Getting Started with Workflow Builder: 5 Workflows We Recommend

- Company: Harvey (harvey.ai)
- Announced: 2025-07-17T15:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/getting-started-with-workflow-builder-5-workflows-we-recommend
- Record: https://forck.live/items/17697-getting-started-with-workflow-builder-5-workflows-we-recommend
- Subject: Harvey

Harvey published a guide recommending five types of workflows to build using its Workflow Builder, which launched to all customers the previous month. The workflows include writing from precedent, generating jurisdictional objections, and comparing draft DPAs against preferred terms. The guide aims to help legal teams create custom workflows that leverage their own expertise and precedents.

### Vercel — Grep a million GitHub repositories via MCP

- Company: Vercel (vercel.com)
- Announced: 2025-07-17T13:00:00+00:00
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/grep-a-million-github-repositories-via-mcp
- Record: https://forck.live/items/20854-grep-a-million-github-repositories-via-mcp
- Subject: v0 / AI SDK / AI Gateway

Vercel announced that Grep now supports the Model Context Protocol (MCP), providing an endpoint that allows AI agents to search over a million public GitHub repositories. The MCP server is built on the same infrastructure as grep.app and returns code snippets matching queries, filtered by language, repository, and file path. Example configurations for Cursor and Claude Code are provided.

### Lovable — Lovable Raises $200M, Valued at $1.8B, Just Eight Months After Launch

- Company: Lovable (lovable.dev)
- Announced: 2025-07-17T10:30:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lovable.dev/blog/200m-series-a-fundraise
- Record: https://forck.live/items/12138-lovable-raises-200m-valued-at-1-8b-just-eight-months-after-launch
- Subject: Lovable / AI app builder

Lovable raised $200 million in a Series A round led by Accel, valuing the company at $1.8 billion, just eight months after its launch. The funding will support its mission to enable non-coders to build production-grade software by conversing with AI.

### Replicate — Bria is now on Replicate

- Company: Replicate (replicate.com)
- Announced: 2025-07-17T10:00:00+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replicate.com/blog/bria
- Record: https://forck.live/items/2484-bria-is-now-on-replicate
- Subject: Platform

Bria's suite of commercial-safe visual AI models is now available on Replicate.

### OpenAI — ChatGPT agent System Card

- Company: OpenAI (openai.com)
- Announced: 2025-07-17T10:00:00+00:00
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/chatgpt-agent-system-card
- Record: https://forck.live/items/526-chatgpt-agent-system-card
- Subject: GPT / ChatGPT / API

OpenAI's ChatGPT agent system card describes an agentic model that combines research, browser automation, and code tools with safeguards.

### OpenAI — Introducing ChatGPT agent

- Company: OpenAI (openai.com)
- Announced: 2025-07-17T10:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/introducing-chatgpt-agent
- Record: https://forck.live/items/525-introducing-chatgpt-agent
- Subject: GPT / ChatGPT / API

OpenAI introduces a ChatGPT agent that can think and act, using tools to complete tasks such as research, bookings, and slideshows with user guidance.

### Supabase — New Observability Features in Supabase

- Company: Supabase (supabase.com)
- Announced: 2025-07-17
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/new-observability-features-in-supabase
- Record: https://forck.live/items/21346-new-observability-features-in-supabase
- Subject: Supabase

Supabase announced new observability features including OpenTelemetry support, a unified logging interface across all services, advanced product reports with metrics like p99 response time for paid customers, and an AI Assistant with debugging capabilities that can retrieve and analyze logs. The new logging interface currently supports API Gateway and Postgres logs, with additional product logs coming soon.

### Supabase — Algolia Connector for Supabase

- Company: Supabase (supabase.com)
- Announced: 2025-07-17
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/algolia-connector-for-supabase
- Record: https://forck.live/items/21345-algolia-connector-for-supabase
- Subject: Supabase

Algolia launched a Supabase Connector that allows developers to index Postgres data from Supabase into Algolia's search service without writing code. The connector syncs selected tables or views on a configurable schedule, and Algolia handles the indexing and search infrastructure.

### Upstage — Upstage Document Intelligence Now Available on the AWS AI Agent Marketplace

- Company: Upstage (upstage.ai)
- Announced: 2025-07-17
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.upstage.ai/blog/en/upstage-document-intelligence-now-available-on-the-aws-ai-agent-marketplace
- Record: https://forck.live/items/18466-upstage-document-intelligence-now-available-on-the-aws-ai-agent-marketplace
- Subject: Solar Pro / Solar Mini

Upstage made Document Intelligence available in the AWS Marketplace's AI Agents & Tools category, offering document parsing and information extraction through a single API call. The service runs serverless inside AWS and can be billed to customers' existing AWS credits.

### Mistral AI — Le Chat dives deep.

- Company: Mistral AI (mistral.ai)
- Announced: 2025-07-17
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://mistral.ai/news/le-chat-dives-deep
- Record: https://forck.live/items/3318-le-chat-dives-deep
- Subject: Mistral / Le Chat

Mistral AI announces a new capability for its Le Chat product, described as 'dives deep'.

### Together AI — Back to The Future: Evaluating AI Agents on Predicting Future Events

- Company: Together AI (together.ai)
- Announced: 2025-07-17
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.together.ai/blog/futurebench
- Record: https://forck.live/items/2422-back-to-the-future-evaluating-ai-agents-on-predicting-future-events
- Subject: Inference platform

Introduces FutureBench, a benchmark for evaluating AI agents on their ability to predict future events, aiming to measure reasoning and synthesis skills while avoiding data contamination by using events that have not yet occurred.

### Together AI — Together AI Delivers Top Speeds for DeepSeek-R1-0528 Inference on NVIDIA Blackwell

- Company: Together AI (together.ai)
- Announced: 2025-07-17
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.together.ai/blog/fastest-inference-for-deepseek-r1-0528-with-nvidia-hgx-b200
- Record: https://forck.live/items/2421-together-ai-delivers-top-speeds-for-deepseek-r1-0528-inference-on-nvidia
- Subject: Inference platform

Together AI announces rolling out NVIDIA Blackwell support for Together Inference, achieving top speeds for DeepSeek-R1-0528 inference with up to 334 tokens/sec.

### Hugging Face — Five Big Improvements to Gradio MCP Servers

- Company: Hugging Face (huggingface.co)
- Announced: 2025-07-17
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/gradio-mcp-updates
- Record: https://forck.live/items/1656-five-big-improvements-to-gradio-mcp-servers
- Subject: Platform

Hugging Face announced five improvements to Gradio MCP servers as of version 5.38.0: seamless local file support via a File Upload MCP server, real-time progress notifications, one-line transformation of OpenAPI specs to MCP, authentication improvements with gr.Header, and customizable tool descriptions with api_description.

### Hugging Face — Back to The Future: Evaluating AI Agents on Predicting Future Events

- Company: Hugging Face (huggingface.co)
- Announced: 2025-07-17
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/futurebench
- Record: https://forck.live/items/1655-back-to-the-future-evaluating-ai-agents-on-predicting-future-events
- Subject: Platform

Hugging Face and collaborators propose FutureBench, a benchmark that evaluates AI agents on their ability to predict future events using real-world prediction markets and news.

### Hugging Face — Consilium: When Multiple LLMs Collaborate

- Company: Hugging Face (huggingface.co)
- Announced: 2025-07-17
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/consilium-multi-llm
- Record: https://forck.live/items/1654-consilium-when-multiple-llms-collaborate
- Subject: Platform

Consilium is a multi-LLM platform that allows multiple AI models to discuss complex questions and reach consensus through structured debate, built during the Gradio Agents & MCP Hackathon.

### OpenAI — Invideo AI uses OpenAI models to create videos 10x faster

- Company: OpenAI (openai.com)
- Announced: 2025-07-17
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/invideo-ai
- Record: https://forck.live/items/529-invideo-ai-uses-openai-models-to-create-videos-10x-faster
- Subject: GPT / ChatGPT / API

Invideo AI uses OpenAI's GPT-4.1, gpt-image-1, and text-to-speech models to create videos.

### OpenAI — Statement from the OpenAI Board of Directors on the Nonprofit Commission Report

- Company: OpenAI (openai.com)
- Announced: 2025-07-17
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/nonprofit-commission-report
- Record: https://forck.live/items/528-statement-from-the-openai-board-of-directors-on-the-nonprofit-commission-report
- Subject: GPT / ChatGPT / API

OpenAI's Board of Directors thanks the independent OpenAI Nonprofit Commission for their extensive work and engagement.

### OpenAI — OpenAI nonprofit jam

- Company: OpenAI (openai.com)
- Announced: 2025-07-17
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/global-affairs/openai-nonprofit-jam
- Record: https://forck.live/items/527-openai-nonprofit-jam
- Subject: GPT / ChatGPT / API

OpenAI Academy, in partnership with Walton Family Foundation and Emerson Collective, is hosting the Nonprofit Jam, a one-day nationwide event for over 1,000 nonprofit leaders across 10 locations.

### Replit — Introducing Domain Purchasing on Replit

- Company: Replit (replit.com)
- Announced: 2025-07-16T22:06:29+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/domain-purchasing-on-replit
- Record: https://forck.live/items/17951-introducing-domain-purchasing-on-replit
- Subject: Replit Agent

Replit added a Domain Purchasing feature that lets users search, buy, and configure custom domains for their apps without leaving the platform. The process includes fast domain search, one-click purchase, and automatic DNS configuration, taking under a minute from search to going live.

### Supabase — Improved Security Controls and A New Home for Security

- Company: Supabase (supabase.com)
- Announced: 2025-07-16T22:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/improved-security-controls
- Record: https://forck.live/items/21347-improved-security-controls-and-a-new-home-for-security
- Subject: Supabase

Supabase launched centralized security documentation, organization-wide MFA enforcement for Pro/Team/Enterprise plans, a private-only Realtime channels setting, and the ability to disable specific Security and Performance Advisor rules. These are the first features in a broader security roadmap that includes YubiKey support, custom roles, and self-service SSO.

### Supabase — Introducing Branching 2.0

- Company: Supabase (supabase.com)
- Announced: 2025-07-16
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/branching-2-0
- Record: https://forck.live/items/21348-introducing-branching-2-0
- Subject: Supabase

Supabase announced Branching 2.0, which removes the previous requirement to connect a GitHub repository to create preview branches. Users can now create branches directly through the dashboard, CLI, or Management API, with or without Git. The update is in early stages and available as a feature preview.

### Hugging Face — Ettin Suite: SoTA Paired Encoders and Decoders

- Company: Hugging Face (huggingface.co)
- Announced: 2025-07-16
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://huggingface.co/blog/ettin
- Record: https://forck.live/items/1657-ettin-suite-sota-paired-encoders-and-decoders
- Subject: Platform

Hugging Face introduces the Ettin Suite, a set of paired encoder-only and decoder-only models (17M-1B parameters) trained on identical data, architecture, and recipes, achieving state-of-the-art performance for open-data models in both categories.

### Supabase — Stripe-To-Postgres Sync Engine as standalone Library

- Company: Supabase (supabase.com)
- Announced: 2025-07-15T22:00:00+00:00
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/stripe-engine-as-sync-library
- Record: https://forck.live/items/21349-stripe-to-postgres-sync-engine-as-standalone-library
- Subject: Supabase

Supabase released stripe-sync-engine as a standalone npm package, @supabase/stripe-sync-engine, allowing developers to integrate it into any backend project including Node.js, Express, or Supabase Edge Functions. Previously only available as a Docker image, the package listens to Stripe webhook events and normalizes them into structured Postgres inserts/updates.

### Supabase — Supabase Analytics Buckets with Iceberg Support

- Company: Supabase (supabase.com)
- Announced: 2025-07-15T17:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/analytics-buckets
- Record: https://forck.live/items/21350-supabase-analytics-buckets-with-iceberg-support
- Subject: Supabase

Supabase launched Analytics Buckets in private alpha, a new storage bucket type optimized for analytics with built-in Apache Iceberg support. The buckets are integrated into Supabase Studio, expose table-level views instead of raw files, and can be queried via the new Supabase Iceberg Wrapper, also in alpha. Alpha limits include two analytics buckets per project, up to five namespaces per bucket, and ten tables per namespace; write capability is still in development.

### Replicate — How we optimized FLUX.1 Kontext [dev]

- Company: Replicate (replicate.com)
- Announced: 2025-07-15T16:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replicate.com/blog/flux-kontext-optimization
- Record: https://forck.live/items/2485-how-we-optimized-flux-1-kontext-dev
- Subject: Platform

Replicate explains how they optimized FLUX.1 Kontext [dev] using TaylorSeer, a method that approximates intermediate diffusion steps by caching derivatives, enabling faster inference without quality loss.

### Mistral AI — Voxtral

- Company: Mistral AI (mistral.ai)
- Announced: 2025-07-15T12:00:00+00:00
- Category: open-weight-release
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://mistral.ai/news/voxtral/
- Record: https://forck.live/items/7363-voxtral
- Subject: Mistral / Le Chat

Mistral AI released Voxtral, a family of open-weight speech understanding models in two sizes (24B and 3B) under the Apache 2.0 license, with capabilities including transcription, Q&A, summarization, multilingual support, and function calling. The models are available for download and via API.

### Supabase — Create a Supabase backend using Figma Make

- Company: Supabase (supabase.com)
- Announced: 2025-07-15
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://supabase.com/blog/figma-make-support-for-supabase
- Record: https://forck.live/items/21351-create-a-supabase-backend-using-figma-make
- Subject: Supabase

With today’s update to Figma Make , you can now build richer prototypes and fully functional web applications , complete with real backend logic using Supabase—all without leaving Figma. This integration brings a faster, more seamless path from idea to reality.

### LG AI Research — Unveiling EXAONE 4.0, the next generation of hybrid AI

- Company: LG AI Research (lgresearch.ai)
- Announced: 2025-07-15
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.lgresearch.ai/blog/view?seq=576
- Record: https://forck.live/items/4783-unveiling-exaone-4-0-the-next-generation-of-hybrid-ai
- Subject: EXAONE

LG AI Research announced EXAONE 4.0, a hybrid AI model that integrates general NLP and reasoning capabilities, available in 32B and 1.2B sizes, and released for research and academic purposes via Hugging Face.

### Z.ai — CogVideoX-3

- Company: Z.ai (z.ai)
- Announced: 2025-07-15
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://docs.z.ai/release-notes/new-released#2025-07-15
- Record: https://forck.live/items/4463-cogvideox-3
- Subject: GLM

Z.ai announced CogVideoX-3, an incremental upgrade to their video generation model with improved quality and new features, including support for start and end frame synthesis.

### Hugging Face — Migrating the Hub from Git LFS to Xet

- Company: Hugging Face (huggingface.co)
- Announced: 2025-07-15
- Category: infrastructure-release
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/migrating-the-hub-to-xet
- Record: https://forck.live/items/1658-migrating-the-hub-from-git-lfs-to-xet
- Subject: Platform

Hugging Face is migrating the Hub's storage backend from Git LFS to Xet, a new storage system that scales with AI workloads. The migration uses a Git LFS Bridge for backward compatibility and background content migrations. Over 500,000 repositories holding 20 PB have moved, and more than 1 million users are now using Xet. The team fixed scaling issues after testing with bartowski's repos.

### OpenAI — Intellectual freedom by design

- Company: OpenAI (openai.com)
- Announced: 2025-07-15
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/global-affairs/intellectual-freedom-by-design
- Record: https://forck.live/items/530-intellectual-freedom-by-design
- Subject: GPT / ChatGPT / API

OpenAI states that ChatGPT is designed to be useful, trustworthy, and adaptable, allowing users to customize it.

### Replit — Introducing Web Search for Replit Agent: Your Smart App Creation Partner

- Company: Replit (replit.com)
- Announced: 2025-07-14T16:16:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/web-search
- Record: https://forck.live/items/17952-introducing-web-search-for-replit-agent-your-smart-app-creation-partner
- Subject: Replit Agent

Replit added web search capabilities to its Replit Agent, enabling it to automatically conduct live internet searches to fill knowledge gaps during app creation. Users can also explicitly ask Agent to perform research. The feature is designed to help Agent understand context such as recent market data, API documentation, and competitor information.

### Supabase — Supabase UI: Platform Kit

- Company: Supabase (supabase.com)
- Announced: 2025-07-14
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/supabase-ui-platform-kit
- Record: https://forck.live/items/21353-supabase-ui-platform-kit
- Subject: Supabase

Supabase released Platform Kit, a set of UI components for the Supabase UI Library designed to help platform builders create platforms on top of Supabase. The components sit on top of the Management API and include database, storage, authentication, and AI components. The library is 100% shadcn/ui compatible and customizable with Tailwind CSS.

### Supabase — Introducing JWT Signing Keys

- Company: Supabase (supabase.com)
- Announced: 2025-07-14
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/jwt-signing-keys
- Record: https://forck.live/items/21352-introducing-jwt-signing-keys
- Subject: Supabase

Supabase announced support for asymmetric JWTs with Supabase Auth, allowing token verification without relying on the Auth server. The company also introduced new publishable and secret API keys to replace the legacy anon and service_role keys. Asymmetric JWT support is available as an opt-in feature, with all new projects defaulting to it from October 1, 2025.

### Cognition — Cognition’s acquisition of Windsurf

- Company: Cognition (cognition.com)
- Announced: 2025-07-14
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cognition.com/blog/windsurf
- Record: https://forck.live/items/7697-cognition-s-acquisition-of-windsurf
- Subject: Devin / SWE

Cognition has signed a definitive agreement to acquire Windsurf, the agentic IDE, including its product, IP, trademark, brand, $82M in ARR, 350+ enterprise customers, and the Windsurf team. In the near term, the Windsurf team will continue to operate as before, with integration of Windsurf's capabilities and IP into Cognition's products planned over the coming months.

### xAI — Announcing xAI for Government

- Company: xAI (x.ai)
- Announced: 2025-07-14
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://x.ai/news/government
- Record: https://forck.live/items/3487-announcing-xai-for-government
- Subject: Grok

xAI announces a suite of frontier AI products available first to United States Government customers.

### Together AI — Kimi K2: Leading Open-Source Model Now Available on Together AI

- Company: Together AI (together.ai)
- Announced: 2025-07-14
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.together.ai/blog/kimi-k2-leading-open-source-model-now-available-on-together-ai
- Record: https://forck.live/items/2423-kimi-k2-leading-open-source-model-now-available-on-together-ai
- Subject: Inference platform

Together AI announces the availability of Kimi K2, a 1 trillion parameter open-source model from Moonshot AI, on its platform with serverless deployment, pricing at $1.00 per 1M input tokens and $3.00 per 1M output tokens, and claims frontier performance across benchmarks.