# forck.live — 8–14 December 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-W50, 8–14 December 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-W50
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### Google Research — Spotlight on innovation: Google-sponsored Data Science for Health Ideathon across Africa

- Company: Google Research (research.google)
- Announced: 2025-12-12T10:42:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://research.google/blog/spotlight-on-innovation-google-sponsored-data-science-for-health-ideathon-across-africa/
- Record: https://forck.live/items/1383-spotlight-on-innovation-google-sponsored-data-science-for-health-ideathon
- Subject: Research

Google Research sponsored an Africa-wide Data Science for Health Ideathon, where teams used Google's open health AI models (MedGemma, TxGemma, MedSigLIP, Gemini) to develop solutions for local healthcare challenges. The post announces the winners and describes their projects.

### Thinking Machines Lab — Tinker: General Availability and Vision Input

- Company: Thinking Machines Lab (thinkingmachines.ai)
- Announced: 2025-12-12
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://thinkingmachines.ai/news/tinker-general-availability
- Record: https://forck.live/items/18561-tinker-general-availability-and-vision-input
- Subject: Thinking Machines / Inkling

Today we are announcing four updates to Tinker No more waitlist New reasoning model: Kimi K2 Thinking New inference interface that is compatible with the OpenAI API Vision input support with Qwen3-VL General availability The waitlist is over!

### Together AI — Announcing Together Python SDK v2.0

- Company: Together AI (together.ai)
- Announced: 2025-12-12
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.together.ai/blog/together-python-sdk-2-0
- Record: https://forck.live/items/2395-announcing-together-python-sdk-v2-0
- Subject: Inference platform

Together AI announces the release candidate of Python SDK v2.0, a modern, strongly-typed, OpenAPI-generated client that replaces the legacy v1 SDK. The new SDK is faster, easier to maintain, and includes beta APIs for Instant Clusters. It features better type safety, httpx-based HTTP client, uv support, and a migration guide with breaking changes.

### OpenAI — BNY builds “AI for everyone, everywhere” with OpenAI

- Company: OpenAI (openai.com)
- Announced: 2025-12-12
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/bny
- Record: https://forck.live/items/339-bny-builds-ai-for-everyone-everywhere-with-openai
- Subject: GPT / ChatGPT / API

BNY is using OpenAI to expand AI adoption across the enterprise via a platform called Eliza, enabling over 20,000 employees to build AI agents that improve efficiency and client outcomes.

### OpenAI — BBVA and OpenAI collaborate to transform global banking

- Company: OpenAI (openai.com)
- Announced: 2025-12-12
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/bbva-collaboration-expansion
- Record: https://forck.live/items/338-bbva-and-openai-collaborate-to-transform-global-banking
- Subject: GPT / ChatGPT / API

BBVA is expanding its collaboration with OpenAI through a multi-year AI transformation program, deploying ChatGPT Enterprise to all 120,000 employees to develop AI solutions for customer interactions and operations.

### OpenAI — How We Used Codex to Ship Sora for Android in 28 Days

- Company: OpenAI (openai.com)
- Announced: 2025-12-12
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/shipping-sora-for-android-with-codex
- Record: https://forck.live/items/337-how-we-used-codex-to-ship-sora-for-android-in-28-days
- Subject: GPT / ChatGPT / API

OpenAI shipped Sora for Android in 28 days using Codex, with AI-assisted planning, translation, and parallel coding workflows.

### Harvey — Building in Active Guardrails With GPT-5.2

- Company: Harvey (harvey.ai)
- Announced: 2025-12-11T18:15:00+00:00
- Category: new-model
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/building-in-guardrails-with-gpt-5-2
- Record: https://forck.live/items/17636-building-in-active-guardrails-with-gpt-5-2
- Subject: Harvey

Harvey announced GPT-5.2, a new model available on its platform that improves capability awareness and proactive guardrails for complex legal work. The model demonstrated 89.8% on Harvey's BigLaw Bench evaluation and is better at knowing its own limitations, surfacing constraints to users rather than failing silently. It will roll out to eligible US clients within 24 hours and later in the EU.

### Runway — Introducing Runway GWM-1

- Company: Runway (runway.com)
- Announced: 2025-12-11T16:22:35.195000+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://runway.com/research/introducing-runway-gwm-1
- Record: https://forck.live/items/7900-introducing-runway-gwm-1
- Subject: Gen / Aleph

Runway announced GWM-1, its first general world model family built on Gen-4.5. The model generates frames autoregressively in real time and can be controlled interactively with actions. It comes in three variants: GWM Worlds for explorable environments, GWM Avatars for conversational characters, and GWM Robotics for robotic manipulation.

### Hugging Face — New in llama.cpp: Model Management

- Company: Hugging Face (huggingface.co)
- Announced: 2025-12-11T15:47:44+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/ggml-org/model-management-in-llamacpp
- Record: https://forck.live/items/1572-new-in-llama-cpp-model-management
- Subject: Platform

llama.cpp server now has a router mode that allows dynamic loading, unloading, and switching between multiple models without restarting the server.

### Harvey — Raising the Bar With Harvey

- Company: Harvey (harvey.ai)
- Announced: 2025-12-11T15:30:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/raising-the-bar-with-harvey
- Record: https://forck.live/items/17637-raising-the-bar-with-harvey
- Subject: Harvey

Harvey announced platform updates that improve performance, personalization, and integration for legal professionals. Key enhancements include up to 14x faster citation latency, the ability to group related files for analysis, image recognition in Assistant, and tailored outputs based on user role and practice area. The updates also introduce a Harvey Guide feature for in-app support and more granular sharing controls for threads and assets.

### OpenAI — Advancing science and math with GPT-5.2

- Company: OpenAI (openai.com)
- Announced: 2025-12-11T10:00:00+00:00
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://openai.com/index/gpt-5-2-for-science-and-math
- Record: https://forck.live/items/340-advancing-science-and-math-with-gpt-5-2
- Subject: GPT / ChatGPT / API

OpenAI announces GPT-5.2, its strongest model for math and science, achieving new state-of-the-art results on benchmarks and enabling real research progress such as solving an open theoretical problem and generating reliable mathematical proofs.

### Google DeepMind — Deepening our partnership with the UK AI Security Institute

- Company: Google DeepMind (deepmind.google)
- Announced: 2025-12-11T00:06:40+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://deepmind.google/blog/deepening-our-partnership-with-the-uk-ai-security-institute/
- Record: https://forck.live/items/1263-deepening-our-partnership-with-the-uk-ai-security-institute
- Subject: Gemini

Google DeepMind and the UK AI Security Institute are strengthening their collaboration on critical AI safety and security research.

### Amp — Thread Map

- Company: Amp (ampcode.com)
- Announced: 2025-12-11
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/thread-map
- Record: https://forck.live/items/7776-thread-map
- Subject: Amp

Amp added a thread map feature to its CLI that shows a top-down view of all threads connected to the current thread via mentions, handoffs, or forks.

### Amp — Thread Labels

- Company: Amp (ampcode.com)
- Announced: 2025-12-11
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/thread-labels
- Record: https://forck.live/items/7775-thread-labels
- Subject: Amp

Amp now allows users to add labels to threads to organize work and find conversations later. Labels can be added via a tag icon, with suggestions from previously used labels, and threads can be filtered by label.

### Amp — Look At This

- Company: Amp (ampcode.com)
- Announced: 2025-12-11
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/look-at
- Record: https://forck.live/items/7774-look-at-this
- Subject: Amp

Amp replaced its look_at tool with a view_media tool. The new tool can return raw or cropped/transformed image data in addition to a textual description, and Amp can now look at PDFs, images, and other media files with a goal in mind.

### Z.ai — AutoGLM-Phone-Multilingual

- Company: Z.ai (z.ai)
- Announced: 2025-12-11
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://docs.z.ai/release-notes/new-released#2025-12-11
- Record: https://forck.live/items/4456-autoglm-phone-multilingual
- Subject: GLM

Z.ai launched AutoGLM-Phone-Multilingual, a multimodal mobile automation framework with ADB support, enabling natural-language task execution across 50+ mainstream apps with multilingual support (English & Chinese), enhanced workflow planning, and improved reliability.

### xAI — xAI and El Salvador Pioneer the World's First Nationwide AI Education Program

- Company: xAI (x.ai)
- Announced: 2025-12-11
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://x.ai/news/el-salvador-partnership
- Record: https://forck.live/items/3481-xai-and-el-salvador-pioneer-the-world-s-first-nationwide-ai-education-program
- Subject: Grok

xAI and the Government of El Salvador announce a partnership to create the world's first nationwide AI education program.

### Hugging Face — Codex is Open Sourcing AI models

- Company: Hugging Face (huggingface.co)
- Announced: 2025-12-11
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/hf-skills-training-codex
- Record: https://forck.live/items/1573-codex-is-open-sourcing-ai-models
- Subject: Platform

Hugging Face published a tutorial showing how to use OpenAI's Codex agent with Hugging Face Skills to perform end-to-end machine learning experiments, including fine-tuning, evaluation, and report generation.

### OpenAI — How Podium is arming 10,000+ SMBs with AI agents

- Company: OpenAI (openai.com)
- Announced: 2025-12-11
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/podium
- Record: https://forck.live/items/345-how-podium-is-arming-10-000-smbs-with-ai-agents
- Subject: GPT / ChatGPT / API

Podium uses OpenAI's GPT-5 to create an AI teammate called Jerry, which helps small and medium-sized businesses and has driven 300% growth.

### OpenAI — The Walt Disney Company and OpenAI reach landmark agreement to bring beloved characters to Sora

- Company: OpenAI (openai.com)
- Announced: 2025-12-11
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/disney-sora-agreement
- Record: https://forck.live/items/344-the-walt-disney-company-and-openai-reach-landmark-agreement-to-bring-beloved
- Subject: GPT / ChatGPT / API

The Walt Disney Company and OpenAI have reached an agreement to bring over 200 Disney, Marvel, Pixar, and Star Wars characters to Sora for fan-inspired short videos, emphasizing responsible AI and including Disney's use of ChatGPT Enterprise and OpenAI API.

### OpenAI — Introducing GPT-5.2

- Company: OpenAI (openai.com)
- Announced: 2025-12-11
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://openai.com/index/introducing-gpt-5-2
- Record: https://forck.live/items/343-introducing-gpt-5-2
- Subject: GPT / ChatGPT / API

OpenAI announces GPT-5.2, a frontier model with advanced reasoning, long-context understanding, coding, and vision capabilities, available in ChatGPT and the OpenAI API.

### OpenAI — Ten years

- Company: OpenAI (openai.com)
- Announced: 2025-12-11
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/ten-years
- Record: https://forck.live/items/342-ten-years
- Subject: GPT / ChatGPT / API

OpenAI publishes a retrospective on its first ten years, highlighting research milestones and the impact of its AI systems, while reaffirming its commitment to building beneficial AGI.

### OpenAI — Update to GPT-5 System Card: GPT-5.2

- Company: OpenAI (openai.com)
- Announced: 2025-12-11
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://openai.com/index/gpt-5-system-card-update-gpt-5-2
- Record: https://forck.live/items/341-update-to-gpt-5-system-card-gpt-5-2
- Subject: GPT / ChatGPT / API

GPT-5.2 is introduced as the latest model family in the GPT-5 series, with a safety mitigation approach largely identical to the previous GPT-5 and GPT-5.1 system cards.

### Google Research — A differentially private framework for gaining insights into AI chatbot use

- Company: Google Research (research.google)
- Announced: 2025-12-10T21:59:41+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/a-differentially-private-framework-for-gaining-insights-into-ai-chatbot-use/
- Record: https://forck.live/items/1384-a-differentially-private-framework-for-gaining-insights-into-ai-chatbot-use
- Subject: Research

Google Research announces a differentially private framework called Urania for gaining insights into AI chatbot usage, presented at COLM 2025. The framework uses DP clustering, DP keyword extraction, and LLM summarization to provide end-to-end privacy guarantees.

### Google DeepMind — Strengthening our partnership with the UK government to support prosperity and security in the AI era

- Company: Google DeepMind (deepmind.google)
- Announced: 2025-12-10T14:59:21+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://deepmind.google/blog/strengthening-our-partnership-with-the-uk-government-to-support-prosperity-and-security-in-the-ai-era/
- Record: https://forck.live/items/1264-strengthening-our-partnership-with-the-uk-government-to-support-prosperity-and
- Subject: Gemini

DeepMind is deepening its partnership with the UK government to support prosperity and security in the AI era.

### Vercel — Vercel launches partner certification

- Company: Vercel (vercel.com)
- Announced: 2025-12-10T13:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://vercel.com/blog/vercel-launches-partner-certification
- Record: https://forck.live/items/20778-vercel-launches-partner-certification
- Subject: v0 / AI SDK / AI Gateway

Vercel announced the inaugural cohort of eleven Certified Solution Partners, a program that validates partner expertise in Next.js and Vercel to help customers select qualified teams for projects such as redesigns, migrations, and new product development.

### Harvey — How Agentic Search Unlocks Legal Research Intelligence at Harvey

- Company: Harvey (harvey.ai)
- Announced: 2025-12-10T13:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/how-agentic-search-unlocks-legal-research-intelligence
- Record: https://forck.live/items/17638-how-agentic-search-unlocks-legal-research-intelligence-at-harvey
- Subject: Harvey

Harvey introduced Agentic Search, a system that moves from static one-shot retrieval to an iterative reasoning process for legal research. The system plans its approach, decides which sources to query, evaluates information sufficiency, and refines its strategy based on discoveries. It serves as the technical foundation for Harvey's Deep Research mode and taps into over 150 legal knowledge sources, internal Vault documents, iManage integration, and LexisNexis case law databases.

### OpenAI — Strengthening cyber resilience as AI capabilities advance

- Company: OpenAI (openai.com)
- Announced: 2025-12-10T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/strengthening-cyber-resilience
- Record: https://forck.live/items/346-strengthening-cyber-resilience-as-ai-capabilities-advance
- Subject: GPT / ChatGPT / API

OpenAI is investing in stronger safeguards and defensive capabilities as AI models become more powerful in cybersecurity, explaining how they assess risk, limit misuse, and work with the security community to strengthen cyber resilience.

### Replit — Building Production-ready Apps with Automated Database Migrations on Replit

- Company: Replit (replit.com)
- Announced: 2025-12-10T01:24:31+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replit.com/blog/production-databases-automated-migrations
- Record: https://forck.live/items/17929-building-production-ready-apps-with-automated-database-migrations-on-replit
- Subject: Replit Agent

Separating development and production databases is fundamental to safe software development - it prevents accidental data loss, allows developers to test changes freely, and ensures production data remains stable and secure.

### Perceptron — Introducing Isaac 0.2

- Company: Perceptron (perceptron.inc)
- Announced: 2025-12-10
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.perceptron.inc/blog/introducing-isaac-0-2
- Record: https://forck.live/items/18619-introducing-isaac-0-2
- Subject: Perceptron Isaac / Perceptron Mk1

### Microsoft AI — It’s About Time: The Copilot Usage Report 2025

- Company: Microsoft AI (microsoft.ai)
- Announced: 2025-12-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://microsoft.ai/news/its-about-time-the-copilot-usage-report-2025
- Record: https://forck.live/items/8435-it-s-about-time-the-copilot-usage-report-2025
- Subject: MAI

Microsoft AI released a usage report analyzing 37.5 million Copilot conversations, finding that health topics dominate mobile usage, programming and gaming crossover on weekdays/weekends, and advice-seeking is on the rise.

### Amp — Amp Python SDK

- Company: Amp (ampcode.com)
- Announced: 2025-12-10
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/python-sdk
- Record: https://forck.live/items/7778-amp-python-sdk
- Subject: Amp

Amp has released a Python SDK for its platform, allowing developers to run Amp programmatically from Python code. The SDK is available via pip as amp-sdk, and requires the Amp CLI installed globally.

### Amp — Agent Skills

- Company: Amp (ampcode.com)
- Announced: 2025-12-10
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/agent-skills
- Record: https://forck.live/items/7777-agent-skills
- Subject: Amp

Amp introduced agent skills, a feature that allows agents to lazily-load instructions for using local tools, installed by default in .agents/skills/ and compatible with existing skills directories.

### Z.ai — GLM-ASR-2512

- Company: Z.ai (z.ai)
- Announced: 2025-12-10
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://docs.z.ai/release-notes/new-released#2025-12-10
- Record: https://forck.live/items/4457-glm-asr-2512
- Subject: GLM

Z.ai launched GLM-ASR-2512, an ASR model with a Character Error Rate of 0.0717 and improved performance in multilingual and accent-rich scenarios, along with enhanced custom dictionary support and expanded specialized terminology recognition.

### Replit — Introducing Race to Revenue to Support the New Generation of Builders

- Company: Replit (replit.com)
- Announced: 2025-12-09T19:51:47+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replit.com/blog/race-to-revenue
- Record: https://forck.live/items/17930-introducing-race-to-revenue-to-support-the-new-generation-of-builders
- Subject: Replit Agent

Replit announced Race to Revenue, an 8-week accelerator program for 20 bootstrapped builders. Participants receive $10,000 in Replit credits, mentorship, and strategic perks from partners such as Stripe Atlas and Rippling, with no equity required. Applications open on December 9, 2025.

### Harvey — The Brief: December 2025

- Company: Harvey (harvey.ai)
- Announced: 2025-12-09T16:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/the-brief-december-2025
- Record: https://forck.live/items/17639-the-brief-december-2025
- Subject: Harvey

Harvey added GPT-5.1 to its model selector across Assistant, Vault, and Workflow Builder, offering stronger reasoning, faster responses, and clearer outputs. The platform also introduced new features including Harvey Guide in Assistant, @ Mention in Composer, integration with Gyldendal Rettsdata for Norwegian legal sources, new local knowledge sources for several jurisdictions, the ability to upload documents alongside Ask LexisNexis, table reading in documents, password-protected file uploads, and workflow enhancements such as embedding vaults and regional knowledge sources and optional steps.

### OpenAI — How Scout24 is building the next generation of real-estate search with AI

- Company: OpenAI (openai.com)
- Announced: 2025-12-09T16:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/scout24
- Record: https://forck.live/items/347-how-scout24-is-building-the-next-generation-of-real-estate-search-with-ai
- Subject: GPT / ChatGPT / API

Scout24 built a conversational assistant powered by GPT-5 that transforms real-estate search by asking clarifying questions, providing summaries, and offering tailored listing recommendations.

### Vercel — Inside Workflow DevKit: How framework integrations work

- Company: Vercel (vercel.com)
- Announced: 2025-12-09T13:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://vercel.com/blog/inside-workflow-devkit-how-framework-integrations-work
- Record: https://forck.live/items/20779-inside-workflow-devkit-how-framework-integrations-work
- Subject: v0 / AI SDK / AI Gateway

Vercel's post explains the two-phase pattern (build-time and runtime) behind the Workflow DevKit (WDK) framework integrations, which now support eight frameworks including SvelteKit, Astro, Express, and Hono, with TanStack Start and React Router in active development. The WDK uses an SWC compiler plugin to transform the same input file into three outputs (client, step, and workflow modes) and handles framework-specific request object conversion via injected helper functions.

### Google DeepMind — FACTS Benchmark Suite: Systematically evaluating the factuality of large language models

- Company: Google DeepMind (deepmind.google)
- Announced: 2025-12-09T11:29:03+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://deepmind.google/blog/facts-benchmark-suite-systematically-evaluating-the-factuality-of-large-language-models/
- Record: https://forck.live/items/1265-facts-benchmark-suite-systematically-evaluating-the-factuality-of-large
- Subject: Gemini

Google DeepMind introduces the FACTS Benchmark Suite, a systematic method for evaluating the factuality of large language models.

### OpenAI — OpenAI co-founds Agentic AI Foundation, donates AGENTS.md

- Company: OpenAI (openai.com)
- Announced: 2025-12-09T09:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/agentic-ai-foundation
- Record: https://forck.live/items/348-openai-co-founds-agentic-ai-foundation-donates-agents-md
- Subject: GPT / ChatGPT / API

OpenAI co-founds the Agentic AI Foundation under the Linux Foundation and donates AGENTS.md to support open, interoperable standards for safe agentic AI.

### OpenAI — Launching our first OpenAI Certifications courses

- Company: OpenAI (openai.com)
- Announced: 2025-12-09T06:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/openai-certificate-courses
- Record: https://forck.live/items/349-launching-our-first-openai-certifications-courses
- Subject: GPT / ChatGPT / API

OpenAI announces new certifications and AI Foundations courses to help people build AI skills and advance careers.

### Amp — 200k Tokens Is Plenty

- Company: Amp (ampcode.com)
- Announced: 2025-12-09
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://ampcode.com/200k-tokens-is-plenty
- Record: https://forck.live/items/7779-200k-tokens-is-plenty
- Subject: Amp

Amp's team argues that a 200k token context window is sufficient for coding, because short threads produce better results and cost less than long ones. The post describes how the team breaks features into small, single-task threads and shares context between them using thread mentions.

### LG AI Research — AI Scientist & AI Victor: Two Geniuses Discuss 'The Future of AI' - Part 1

- Company: LG AI Research (lgresearch.ai)
- Announced: 2025-12-09
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.lgresearch.ai/blog/view?seq=605
- Record: https://forck.live/items/4769-ai-scientist-ai-victor-two-geniuses-discuss-the-future-of-ai-part-1
- Subject: EXAONE

A blog post featuring a dialogue between LG AI Research's Chief Scientist Honglak Lee and Go legend Sedol Lee, discussing the impact of AlphaGo, the current state of AI technology, and South Korea's AI capabilities and challenges.

### Baidu — ERNIE-5.0-Preview-1103 landed on the LMArena Text Leaderboard!

- Company: Baidu (baidu.com)
- Announced: 2025-12-09
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://ernie.baidu.com/blog/posts/ernie-5.0-preview-1103-release-on-lmarena/
- Record: https://forck.live/items/4648-ernie-5-0-preview-1103-landed-on-the-lmarena-text-leaderboard
- Subject: ERNIE

Baidu announced the release of ERNIE-5.0-Preview-1103 on the LMArena leaderboard, where it ranks in the top 20. The model features omni-modal modeling, unified understanding and generation, a unified autoregressive architecture, a massive-scale MoE architecture with over 2 trillion parameters, and enhanced agentic capabilities for long-horizon tasks. It achieves state-of-the-art performance on benchmarks, particularly in multimodal understanding, instruction following, creative writing, factual reasoning, agentic planning, and tool use.

### Mistral AI — Introducing: Devstral 2 and Mistral Vibe CLI.

- Company: Mistral AI (mistral.ai)
- Announced: 2025-12-09
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://mistral.ai/news/devstral-2-vibe-cli
- Record: https://forck.live/items/3308-introducing-devstral-2-and-mistral-vibe-cli
- Subject: Mistral / Le Chat

Mistral AI announces Devstral 2 and Mistral Vibe CLI.

### OpenAI — OpenAI appoints Denise Dresser as Chief Revenue Officer

- Company: OpenAI (openai.com)
- Announced: 2025-12-09
- Category: funding-company-news
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/openai-appoints-denise-dresser
- Record: https://forck.live/items/352-openai-appoints-denise-dresser-as-chief-revenue-officer
- Subject: GPT / ChatGPT / API

Denise Dresser has been appointed as Chief Revenue Officer at OpenAI, responsible for global revenue strategy and customer success.

### OpenAI — Commonwealth Bank of Australia builds AI fluency at scale

- Company: OpenAI (openai.com)
- Announced: 2025-12-09
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/commonwealth-bank-of-australia
- Record: https://forck.live/items/351-commonwealth-bank-of-australia-builds-ai-fluency-at-scale
- Subject: GPT / ChatGPT / API

Commonwealth Bank of Australia partners with OpenAI to deploy ChatGPT Enterprise to 50,000 employees to improve customer service and fraud response.

### OpenAI — Bringing powerful AI to millions across Europe with Deutsche Telekom

- Company: OpenAI (openai.com)
- Announced: 2025-12-09
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/deutsche-telekom-collaboration
- Record: https://forck.live/items/350-bringing-powerful-ai-to-millions-across-europe-with-deutsche-telekom
- Subject: GPT / ChatGPT / API

OpenAI is collaborating with Deutsche Telekom to bring advanced, multilingual AI experiences to millions across Europe and deploy ChatGPT Enterprise for employees.

### Alibaba — Qwen3-Omni-Flash-2025-12-01：Hear You. See You. Follow Smarter!

- Company: Alibaba (alibaba.com)
- Announced: 2025-12-08T21:00:00+00:00
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://qwen.ai/blog?id=qwen3-omni-flash-20251201
- Record: https://forck.live/items/4593-qwen3-omni-flash-2025-12-01-hear-you-see-you-follow-smarter
- Subject: Qwen

Qwen3-Omni-Flash-2025-12-01 is an upgraded version of Qwen3-Omni, a native multimodal model that processes text, images, audio, and video, and outputs text and speech simultaneously. This version introduces enhancements in audio-visual interaction, system prompt control, multilingual capabilities, and speech quality.

### Harvey — Harvey’s Integration With NetDocuments Intelligent DMS

- Company: Harvey (harvey.ai)
- Announced: 2025-12-08T18:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/harvey-netdocuments-integration
- Record: https://forck.live/items/17640-harvey-s-integration-with-netdocuments-intelligent-dms
- Subject: Harvey

Harvey announced an integration with NetDocuments' intelligent DMS, allowing legal teams to securely access and use NetDocuments files directly within Harvey without creating local copies or altering existing permissions. The integration is currently available in early access for Harvey customers.

### OpenAI — Instacart and OpenAI partner on AI shopping experiences

- Company: OpenAI (openai.com)
- Announced: 2025-12-08T06:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/instacart-partnership
- Record: https://forck.live/items/353-instacart-and-openai-partner-on-ai-shopping-experiences
- Subject: GPT / ChatGPT / API

OpenAI and Instacart are deepening their partnership to bring the first fully integrated grocery shopping and Instant Checkout payment app to ChatGPT.

### OpenAI — The state of enterprise AI

- Company: OpenAI (openai.com)
- Announced: 2025-12-08T04:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/the-state-of-enterprise-ai-2025-report
- Record: https://forck.live/items/354-the-state-of-enterprise-ai
- Subject: GPT / ChatGPT / API

OpenAI's enterprise data indicates accelerating AI adoption, deeper integration, and measurable productivity gains across industries in 2025.

### Supabase — Introducing iceberg-js: A JavaScript Client for Apache Iceberg

- Company: Supabase (supabase.com)
- Announced: 2025-12-08
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/introducing-iceberg-js
- Record: https://forck.live/items/21314-introducing-iceberg-js-a-javascript-client-for-apache-iceberg
- Subject: Supabase

Supabase released iceberg-js, a minimal, vendor-agnostic JavaScript client for the Apache Iceberg REST Catalog API. The library provides a 1:1 mapping to the Iceberg REST API for managing namespaces and tables, but does not support data operations, query execution, or engine-specific integrations. It is MIT licensed and works in Node.js, TypeScript, browsers, Deno, and major bundlers.

### Amp — Find Threads

- Company: Amp (ampcode.com)
- Announced: 2025-12-08
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/find-threads
- Record: https://forck.live/items/7780-find-threads
- Subject: Amp

Amp introduced a new find_thread tool that allows users to search threads by keyword or by file, enabling users to find threads that mention specific terms or have touched a specific file. The tool complements the existing read_thread capability.

### Z.ai — GLM-4.6V

- Company: Z.ai (z.ai)
- Announced: 2025-12-08
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://docs.z.ai/release-notes/new-released#2025-12-08
- Record: https://forck.live/items/4458-glm-4-6v
- Subject: GLM

Z.ai released GLM-4.6V, a new version of its multimodal model with enhanced vision understanding and a 128K context window.

### OpenAI — How Virgin Atlantic uses AI to enhance every step of travel

- Company: OpenAI (openai.com)
- Announced: 2025-12-08
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/virgin-atlantic-oliver-byers
- Record: https://forck.live/items/355-how-virgin-atlantic-uses-ai-to-enhance-every-step-of-travel
- Subject: GPT / ChatGPT / API

Virgin Atlantic's CFO describes how the airline uses AI to accelerate development, enhance decision-making, and improve customer experience.