# forck.live — 17–23 November 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-W47, 17–23 November 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-W47
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### Replit — Campus x Replit: Vibe Coding 101 Course

- Company: Replit (replit.com)
- Announced: 2025-11-23T22:43:02+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replit.com/blog/campus-vibe-coding-101-course
- Record: https://forck.live/items/17933-campus-x-replit-vibe-coding-101-course
- Subject: Replit Agent

Replit and Campus launched Vibe Coding 101, an 8-week live online course for beginners to design, build, and deploy apps using AI tools. The course requires no technical background and aims to help learners ship real projects from week one. Registration is open with limited seats, and the first cohort starts in January.

### Cua — Computer-Use Agents for Growth Hacking

- Company: Cua (cua.ai)
- Announced: 2025-11-22
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://cua.ai/blog/computer-use-agents-for-growth-hacking
- Record: https://forck.live/items/16826-computer-use-agents-for-growth-hacking
- Subject: Cua Driver / Lume / Cua Bench / Cua Fleets

Cua describes a guerrilla marketing campaign at Google DevFest Toronto using $200 worth of branded keychains and a post-event automation pipeline built with its computer-use agent. The agent extracts LinkedIn connections and generates personalized follow-up messages, and the company has open-sourced the automation as a cookbook.

### Google Research — Reducing EV range anxiety: How a simple AI model predicts port availability

- Company: Google Research (research.google)
- Announced: 2025-11-21T16:27:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/reducing-ev-range-anxiety-how-a-simple-ai-model-predicts-port-availability/
- Record: https://forck.live/items/1387-reducing-ev-range-anxiety-how-a-simple-ai-model-predicts-port-availability
- Subject: Research

Google Research developed a simple linear regression model to predict the probability of EV charging port availability at a station within a given time frame, using hour-of-day features. The model outperforms a baseline of assuming no change in port availability.

### Harvey — Contract Intelligence: A Scaled Benchmark for Measuring Contract Understanding

- Company: Harvey (harvey.ai)
- Announced: 2025-11-21T16:00:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/contract-intelligence-benchmark
- Record: https://forck.live/items/17645-contract-intelligence-a-scaled-benchmark-for-measuring-contract-understanding
- Subject: Harvey

Harvey published initial findings from its Contract Intelligence benchmark, which measures LLM contract understanding across over 4,000 data points. The results show that lawyers using Review tables in Vault outperformed either lawyers or LLMs alone by 5% or more.

### Vercel — Self-driving infrastructure

- Company: Vercel (vercel.com)
- Announced: 2025-11-21T13:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/self-driving-infrastructure
- Record: https://forck.live/items/20788-self-driving-infrastructure
- Subject: v0 / AI SDK / AI Gateway

Vercel announced Vercel Agent, a system that autonomously monitors deployments, investigates anomalies, runs root-cause analyses, and performs AI code review before deployment. The agent currently requires human approval for recommendations, with plans for gradual autonomy. Vercel also described its broader vision of self-driving infrastructure that closes the feedback loop between production data and code.

### Vercel — Vercel Open Source Program: Fall 2025 cohort

- Company: Vercel (vercel.com)
- Announced: 2025-11-21T13:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/vercel-open-source-program-fall-2025-cohort
- Record: https://forck.live/items/20787-vercel-open-source-program-fall-2025-cohort
- Subject: v0 / AI SDK / AI Gateway

In April, we launched the Vercel Open Source Program to give maintainers the resources, credits, and support they need to ship faster and scale confidently. The first group joined through our spring 2025 cohort .

### ElevenLabs — We’re partnering with Liberty Global to accelerate voice AI expansion across Europe

- Company: ElevenLabs (elevenlabs.io)
- Announced: 2025-11-21T12:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://elevenlabs.io/blog/liberty-global-partnership
- Record: https://forck.live/items/3375-we-re-partnering-with-liberty-global-to-accelerate-voice-ai-expansion-across
- Subject: Voice AI

ElevenLabs is partnering with Liberty Global to accelerate voice AI expansion across Europe.

### Upstage — Upstage × Karakuri co-developed “Syn Pro” officially certified as a domestic foundation model by Japan’s METI

- Company: Upstage (upstage.ai)
- Announced: 2025-11-21
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.upstage.ai/news/syn-pro-japan-fm
- Record: https://forck.live/items/18426-upstage-karakuri-co-developed-syn-pro-officially-certified-as-a-domestic
- Title in English (translated by forck.live from ja): Upstage × Karakuri co-developed “Syn Pro” officially certified as a domestic foundation model by Japan’s METI
- Subject: Solar Pro / Solar Mini

Upstage AI announced that Syn Pro, a Japanese-language LLM co-developed with Karakuri, has been officially certified as a 'Domestic Foundation Model' under Japan's GENIAC program led by METI and NEDO. The certification is granted to models developed within Japan and recognizes Syn Pro's Japanese accuracy, safety, and on-premise deployment support. Syn Pro is a 31B-parameter model that ranked in the global Top 20 on the Nejumi Leaderboard as of November 20, 2025.

### Baidu — ERNIE-5.0-Preview-1120, ready for testing in LMArena!

- Company: Baidu (baidu.com)
- Announced: 2025-11-21
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://ernie.baidu.com/blog/posts/ernie-5.0-preview-1120-release-on-lmarena/
- Record: https://forck.live/items/4649-ernie-5-0-preview-1120-ready-for-testing-in-lmarena
- Subject: ERNIE

ERNIE-5.0-Preview-1120 has been released and entered LMArena Vision Leaderboard for the first time, achieving a score of 1206, ranking in the top 15, top 1 among domestic models, and on par with Claude Sonnet 4 and GPT-5-high. It is natively multimodal.

### Hugging Face — Open ASR Leaderboard: Trends and Insights with New Multilingual & Long-Form Tracks

- Company: Hugging Face (huggingface.co)
- Announced: 2025-11-21
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/open-asr-leaderboard
- Record: https://forck.live/items/1583-open-asr-leaderboard-trends-and-insights-with-new-multilingual-long-form-tracks
- Subject: Platform

The Open ASR Leaderboard has added multilingual and long-form transcription tracks. A preprint analyzing 60+ models from 18 organizations across 11 datasets reveals that Conformer encoder + LLM decoder models achieve the best accuracy, CTC/TDT decoders are the fastest, multilingual models sacrifice single-language performance, and closed-source systems still lead in long-form transcription.

### Hugging Face — 20x Faster TRL Fine-tuning with RapidFire AI

- Company: Hugging Face (huggingface.co)
- Announced: 2025-11-21
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/rapidfireai
- Record: https://forck.live/items/1582-20x-faster-trl-fine-tuning-with-rapidfire-ai
- Subject: Platform

Hugging Face announces official integration of RapidFire AI with TRL, allowing users to launch multiple fine-tuning configurations concurrently with adaptive chunk-based scheduling, achieving up to 16-24x higher experimentation throughput. The tool provides drop-in wrappers, a dashboard for live metrics and control, and multi-GPU orchestration.

### Google Antigravity — Nano Banana Pro in Google Antigravity

- Company: Google Antigravity (antigravity.google)
- Announced: 2025-11-20T18:19:47+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://antigravity.google/blog/nano-banana-pro
- Record: https://forck.live/items/4559-nano-banana-pro-in-google-antigravity
- Subject: Antigravity

Nano Banana Pro, a new image generation model based on Gemini 3 Pro, is rolling out to Google Antigravity users. It offers improved text rendering, edit consistency, and collateral generation compared to the original Nano Banana.

### OpenAI — OpenAI and Foxconn collaborate to strengthen U.S. manufacturing across the AI supply chain

- Company: OpenAI (openai.com)
- Announced: 2025-11-20T14:50:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/openai-and-foxconn-collaborate
- Record: https://forck.live/items/370-openai-and-foxconn-collaborate-to-strengthen-u-s-manufacturing-across-the-ai
- Subject: GPT / ChatGPT / API

OpenAI and Foxconn announced a collaboration to design and manufacture next-generation AI infrastructure hardware in the U.S., including multiple generations of data-center systems, to strengthen U.S. supply chains and build key components domestically.

### Harvey — Resolve Complex Work Faster With Email Harvey

- Company: Harvey (harvey.ai)
- Announced: 2025-11-20T13:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/email-harvey
- Record: https://forck.live/items/17646-resolve-complex-work-faster-with-email-harvey
- Subject: Harvey

Harvey announced Email Harvey, a new feature that lets legal teams send questions or forward email threads to ask@harvey.ai and receive replies, summaries, edits, and drafts directly in their inbox. The feature is designed to help lawyers work on the go and continue conversations in Harvey's web app. Email Harvey is an upcoming feature for current Harvey customers.

### OpenAI — Helping 1,000 small businesses build with AI

- Company: OpenAI (openai.com)
- Announced: 2025-11-20T06:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/small-business-ai-jam
- Record: https://forck.live/items/371-helping-1-000-small-businesses-build-with-ai
- Subject: GPT / ChatGPT / API

OpenAI is partnering with DoorDash, SCORE, and local organizations to help 1,000 small businesses build with AI through the Small Business AI Jam program.

### Replit — Data Connectors

- Company: Replit (replit.com)
- Announced: 2025-11-20T00:46:38+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/data-connectors
- Record: https://forck.live/items/17932-data-connectors
- Subject: Replit Agent

Replit has introduced native data warehouse connectors for Snowflake, Databricks, and BigQuery, available to all Replit Enterprise customers. The connectors use OAuth authentication and allow the Replit Agent to directly query data warehouses, explore schemas, and build applications with live data, eliminating the need for manual credential management.

### Genspark — Genspark AI Workspace: Where Your Work Gets Finished Autonomously

- Company: Genspark (genspark.ai)
- Announced: 2025-11-20
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.genspark.ai/blog/genspark-ai-workspace-and-series-b-funding
- Record: https://forck.live/items/18263-genspark-ai-workspace-where-your-work-gets-finished-autonomously
- Subject: Genspark

### Factory AI — The Droid Wars: Breaking up an AI‑orchestrated cyber fraud campaign

- Company: Factory AI (factory.com)
- Announced: 2025-11-20
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://factory.com/news/droid-neutralizing-fraud
- Record: https://forck.live/items/17834-the-droid-wars-breaking-up-an-ai-orchestrated-cyber-fraud-campaign
- Subject: Droid

Factory AI detected and disrupted a highly automated cyber fraud campaign in October 2025. The attackers used AI coding agents to generate infrastructure, adapt to defenses, and orchestrate traffic from tens of thousands of synthetic organizations, aiming to exploit free compute at scale. Factory AI assesses with high confidence that the attack was conducted by a large, China-based operation, with at least one state-linked actor involved.

### Cua — Cua Playground: Agents + Sandboxes in Your Browser

- Company: Cua (cua.ai)
- Announced: 2025-11-20
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cua.ai/blog/cua-playground-preview
- Record: https://forck.live/items/16827-cua-playground-agents-sandboxes-in-your-browser
- Subject: Cua Driver / Lume / Cua Bench / Cua Fleets

Cua launched a browser-based Playground for testing computer-use agents without writing code. Users can send messages to running sandboxes, watch agents execute in real time, and iterate on prompts, with support for Claude Sonnet 4.5 and Haiku 4.5.

### Replicate — How to prompt Nano Banana Pro

- Company: Replicate (replicate.com)
- Announced: 2025-11-20
- Category: capability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replicate.com/blog/how-to-prompt-nano-banana-pro
- Record: https://forck.live/items/2471-how-to-prompt-nano-banana-pro
- Subject: Platform

Advanced capabilities of Nano Banana Pro, including logic reasoning from images, text adherence, code rendering, and character consistency across multiple reference images.

### Hugging Face — Introducing AnyLanguageModel: One API for Local and Remote LLMs on Apple Platforms

- Company: Hugging Face (huggingface.co)
- Announced: 2025-11-20
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/anylanguagemodel
- Record: https://forck.live/items/1584-introducing-anylanguagemodel-one-api-for-local-and-remote-llms-on-apple
- Subject: Platform

Hugging Face introduced AnyLanguageModel, a Swift package that provides a unified API for local and remote LLMs on Apple platforms, supporting multiple providers including Apple Foundation Models, Core ML, MLX, llama.cpp, Ollama, OpenAI, Anthropic, Google Gemini, and Hugging Face Inference Providers.

### OpenAI — Early experiments in accelerating science with GPT-5

- Company: OpenAI (openai.com)
- Announced: 2025-11-20
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/accelerating-science-gpt-5
- Record: https://forck.live/items/372-early-experiments-in-accelerating-science-with-gpt-5
- Subject: GPT / ChatGPT / API

OpenAI presents initial research studies demonstrating how GPT-5 speeds up scientific progress in fields such as mathematics, physics, biology, and computer science.

### Replit — Fast Mode

- Company: Replit (replit.com)
- Announced: 2025-11-19T17:12:14+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/fast-mode
- Record: https://forck.live/items/17934-fast-mode
- Subject: Replit Agent

Replit launched Fast Mode, a new feature for its Agent that enables quick, targeted adjustments with an efficient model that is more affordable than Build mode. Fast Mode is available for free for a limited time to all Replit users starting today.

### Stability AI — Warner Music Group and Stability AI Join Forces To Build The Next Generation Of Responsible AI Tools For Music Creation

- Company: Stability AI (stability.ai)
- Announced: 2025-11-19T16:01:30+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://stability.ai/news-updates/warner-music-group-and-stability-ai-join-forces-to-build-next-gen-tools
- Record: https://forck.live/items/17478-warner-music-group-and-stability-ai-join-forces-to-build-the-next-generation
- Subject: Stable Diffusion

Warner Music Group and Stability AI announced a collaboration to develop professional-grade AI tools for music creation, focusing on ethically trained models that protect creators' rights and enable new revenue streams.

### Mistral AI — Mistral - KI für Deutschland

- Company: Mistral AI (mistral.ai)
- Announced: 2025-11-19T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://mistral.ai/news/ki-fur-deutschland/
- Record: https://forck.live/items/19884-mistral-ki-f-r-deutschland
- Subject: Mistral / Le Chat

Company Mistral AI - KI für Deutschland November 19, 2025 By Mistral AI

### OpenAI — Strengthening our safety ecosystem with external testing

- Company: OpenAI (openai.com)
- Announced: 2025-11-19T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/strengthening-safety-with-external-testing
- Record: https://forck.live/items/373-strengthening-our-safety-ecosystem-with-external-testing
- Subject: GPT / ChatGPT / API

OpenAI discusses working with independent experts to evaluate frontier AI systems through third-party testing to strengthen safety, validate safeguards, and increase transparency.

### OpenAI — How evals drive the next chapter in AI for businesses

- Company: OpenAI (openai.com)
- Announced: 2025-11-19T11:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/evals-drive-next-chapter-of-ai
- Record: https://forck.live/items/374-how-evals-drive-the-next-chapter-in-ai-for-businesses
- Subject: GPT / ChatGPT / API

Evals help businesses define, measure, and improve AI performance, reducing risk and boosting productivity.

### Google Research — Real-time speech-to-speech translation

- Company: Google Research (research.google)
- Announced: 2025-11-19T09:59:00+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://research.google/blog/real-time-speech-to-speech-translation/
- Record: https://forck.live/items/1388-real-time-speech-to-speech-translation
- Subject: Research

Google Research introduces an end-to-end speech-to-speech translation model that enables real-time translation in the original speaker's voice with only a 2-second delay, using a streaming architecture and a scalable data acquisition pipeline.

### OpenAI — OpenAI and Target team up on new AI-powered experiences

- Company: OpenAI (openai.com)
- Announced: 2025-11-19T06:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/target-partnership
- Record: https://forck.live/items/375-openai-and-target-team-up-on-new-ai-powered-experiences
- Subject: GPT / ChatGPT / API

OpenAI announces a partnership with Target to integrate a Target app into ChatGPT and expand Target's use of ChatGPT Enterprise.

### Hugging Face — Apriel-H1: The Surprising Key to Distilling Efficient Reasoning Models

- Company: Hugging Face (huggingface.co)
- Announced: 2025-11-19T05:19:07+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://huggingface.co/blog/ServiceNow-AI/apriel-h1
- Record: https://forck.live/items/1585-apriel-h1-the-surprising-key-to-distilling-efficient-reasoning-models
- Subject: Platform

ServiceNow AI converted their 15B reasoning model to a Mamba hybrid (Apriel-H1-15b-Thinker-SFT) achieving 2.1x throughput with minimal quality loss, using reverse KL divergence and staged distillation on high-quality reasoning traces.

### Cua — Cua VLM Router: One Provider for All Your Computer-Use Models

- Company: Cua (cua.ai)
- Announced: 2025-11-19
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cua.ai/blog/cua-vlm-router
- Record: https://forck.live/items/16828-cua-vlm-router-one-provider-for-all-your-computer-use-models
- Subject: Cua Driver / Lume / Cua Bench / Cua Fleets

Cua launched the Cua VLM Router, a managed inference API providing unified access to multiple vision-language model providers through a single API key. The initial release supports Anthropic's Claude Sonnet 4.5 and Haiku 4.5 models, with automatic routing to the best available provider among Anthropic, AWS Bedrock, or Microsoft Foundry. Requests are billed in credits deducted from the Cua account balance, with each response including both the gateway cost and the upstream API cost.

### Suno — The Future of Music is Already Here

- Company: Suno (suno.com)
- Announced: 2025-11-19
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://suno.com/blog/series-c-announcement
- Record: https://forck.live/items/10329-the-future-of-music-is-already-here
- Subject: Music / Studio

Suno announced a $250M Series C funding round at a $2.45B post-money valuation, led by Menlo Ventures with participation from NVentures, Hallwood Media, Lightspeed, and Matrix. The company reports that nearly 100 million people have used Suno to create music over the past two years.

### Luma — AGI is multimodal and reality is the dataset of AGI

- Company: Luma (lumalabs.ai)
- Announced: 2025-11-19
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lumalabs.ai/news/series-c
- Record: https://forck.live/items/7977-agi-is-multimodal-and-reality-is-the-dataset-of-agi
- Subject: Ray / Dream Machine

Luma announced a $900 million Series C funding round led by Humain, with significant participation from AMD, and plans to build a 2GW compute supercluster called Project Halo. The company also highlighted its Ray3 reasoning video model and reiterated its vision for multimodal AGI.

### xAI — Grok goes Global with KSA

- Company: xAI (x.ai)
- Announced: 2025-11-19
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://x.ai/news/grok-goes-global
- Record: https://forck.live/items/3483-grok-goes-global-with-ksa
- Subject: Grok

xAI announces a landmark partnership with Saudi Arabia and HUMAIN, expanding Grok's global availability.

### xAI — Grok 4.1 Fast and Agent Tools API

- Company: xAI (x.ai)
- Announced: 2025-11-19
- Category: api-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://x.ai/news/grok-4-1-fast
- Record: https://forck.live/items/3482-grok-4-1-fast-and-agent-tools-api
- Subject: Grok

xAI announces Grok 4.1 Fast and Agent Tools API, bringing next-generation tool-calling agents to the xAI API.

### Mistral AI — Mistral AI - KI für Deutschland

- Company: Mistral AI (mistral.ai)
- Announced: 2025-11-19
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://mistral.ai/news/ki-fur-deutschland
- Record: https://forck.live/items/3310-mistral-ai-ki-f-r-deutschland
- Subject: Mistral / Le Chat

Mistral AI published a post on November 19, 2025, titled 'Mistral AI - KI für Deutschland'.

### Replicate — Retro Diffusion's pixel art models are now on Replicate

- Company: Replicate (replicate.com)
- Announced: 2025-11-19
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replicate.com/blog/retro-diffusions-pixel-art-models-are-now-on-replicate
- Record: https://forck.live/items/2472-retro-diffusion-s-pixel-art-models-are-now-on-replicate
- Subject: Platform

Retro Diffusion's pixel art models are now available on the Replicate platform, allowing users to generate pixel art images, tiles, and animations through Replicate's API.

### OpenAI — How Scania accelerates work with AI across its global workforce

- Company: OpenAI (openai.com)
- Announced: 2025-11-19
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/scania
- Record: https://forck.live/items/379-how-scania-accelerates-work-with-ai-across-its-global-workforce
- Subject: GPT / ChatGPT / API

Scania, a global manufacturer, is scaling AI using ChatGPT Enterprise, with team-based onboarding and guardrails, leading to improvements in productivity, quality, and innovation.

### OpenAI — A free version of ChatGPT built for teachers

- Company: OpenAI (openai.com)
- Announced: 2025-11-19
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/chatgpt-for-teachers
- Record: https://forck.live/items/378-a-free-version-of-chatgpt-built-for-teachers
- Subject: GPT / ChatGPT / API

OpenAI offers ChatGPT for Teachers, a secure workspace with education-grade privacy and admin controls, free for verified U.S. K-12 educators until June 2027.

### OpenAI — Building more with GPT-5.1-Codex-Max

- Company: OpenAI (openai.com)
- Announced: 2025-11-19
- Category: new-model
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/gpt-5-1-codex-max
- Record: https://forck.live/items/377-building-more-with-gpt-5-1-codex-max
- Subject: GPT / ChatGPT / API

OpenAI announced GPT-5.1-Codex-Max, a new coding model for Codex with improved speed, intelligence, reasoning, and token efficiency.

### OpenAI — GPT-5.1-Codex-Max System Card

- Company: OpenAI (openai.com)
- Announced: 2025-11-19
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://openai.com/index/gpt-5-1-codex-max-system-card
- Record: https://forck.live/items/376-gpt-5-1-codex-max-system-card
- Subject: GPT / ChatGPT / API

OpenAI published a system card for GPT‑5.1-CodexMax, describing safety measures including model-level mitigations and product-level mitigations.

### Replit — Design Mode

- Company: Replit (replit.com)
- Announced: 2025-11-18T18:23:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/design-mode
- Record: https://forck.live/items/17927-design-mode
- Subject: Replit Agent

Replit launched Design Mode, a new interface powered by the Gemini 3 model that lets users create interactive mockups and static websites using natural language. Designs start as front-end only, with an option to convert them into full applications. The feature is available on replit.com.

### Google Antigravity — Introducing Google Antigravity, a New Era in AI-Assisted Software Development

- Company: Google Antigravity (antigravity.google)
- Announced: 2025-11-18T18:19:47+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://antigravity.google/blog/introducing-google-antigravity
- Record: https://forck.live/items/4560-introducing-google-antigravity-a-new-era-in-ai-assisted-software-development
- Subject: Antigravity

Google Antigravity announces its new agentic development platform, available in public preview at no charge with rate limits on Gemini 3 Pro usage. The platform aims to enable autonomous, end-to-end software tasks with features for trust, autonomy, feedback, and self-improvement.

### Harvey — Gemini 3 Pro Public Preview: Results From Early Access Evaluations

- Company: Harvey (harvey.ai)
- Announced: 2025-11-18T17:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/gemini-3-pro-public-preview-early-access-evaluation-results
- Record: https://forck.live/items/17647-gemini-3-pro-public-preview-results-from-early-access-evaluations
- Subject: Harvey

Harvey tested Google's Gemini 3 Pro in early access and reports it scored 87.9% on BigLaw Bench, up from Gemini 2.5 Pro's 85%. The model showed stronger reasoning, cleaner structure, and better tone control in legal drafting tasks. Harvey is working with Google and DeepMind to bring the model to general availability within its platform.

### Google DeepMind — We’re expanding our presence in Singapore to advance AI in the Asia-Pacific region

- Company: Google DeepMind (deepmind.google)
- Announced: 2025-11-18T17:00:00+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://deepmind.google/blog/were-expanding-our-presence-in-singapore-to-advance-ai-in-the-asia-pacific-region/
- Record: https://forck.live/items/1274-we-re-expanding-our-presence-in-singapore-to-advance-ai-in-the-asia-pacific
- Subject: Gemini

Google DeepMind has opened a new research lab in Singapore to accelerate AI progress in the Asia-Pacific region.

### Harvey — Harvey’s Improved Word Experience Understands How You Work

- Company: Harvey (harvey.ai)
- Announced: 2025-11-18T16:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/improved-word-experience
- Record: https://forck.live/items/17648-harvey-s-improved-word-experience-understands-how-you-work
- Subject: Harvey

Harvey announced enhancements to its Word Add-In for legal professionals, including a redesigned homepage with prompt suggestions, improved playbook review with granular controls and insights, and new workflows for redacting sensitive information, converting documents to templates, and reviewing definitions. The update also integrates 80+ region-specific knowledge sources for grounding results. Features like specifying party representation and required language flags are planned for December and early next year.

### Lovable — One year of Lovable: Welcome to the age of the builder

- Company: Lovable (lovable.dev)
- Announced: 2025-11-18T16:00:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lovable.dev/blog/one-year-of-lovable
- Record: https://forck.live/items/12119-one-year-of-lovable-welcome-to-the-age-of-the-builder
- Subject: Lovable / AI app builder

Lovable reported that its annualized revenue reached $200 million, doubling in four months, and that its platform now sees 5 million daily visits to user-built sites and apps. The company also announced new MCP server integrations with tools like Linear and Notion, design themes, and plans to open offices in Boston and San Francisco.

### Google Research — Generative UI: A rich, custom, visual interactive user experience for any prompt

- Company: Google Research (research.google)
- Announced: 2025-11-18T16:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/generative-ui-a-rich-custom-visual-interactive-user-experience-for-any-prompt/
- Record: https://forck.live/items/1389-generative-ui-a-rich-custom-visual-interactive-user-experience-for-any-prompt
- Subject: Research

Google Research introduces a novel implementation of generative UI that dynamically creates immersive visual experiences and interactive interfaces. The capability is rolling out in the Gemini app (as dynamic view) and in Google Search (starting with AI Mode), powered by Gemini 3 Pro.

### Lovable — What's new in Lovable: MCP servers and more design power

- Company: Lovable (lovable.dev)
- Announced: 2025-11-18T15:30:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lovable.dev/blog/mcp-servers
- Record: https://forck.live/items/12120-what-s-new-in-lovable-mcp-servers-and-more-design-power
- Subject: Lovable / AI app builder

Lovable released MCP-based integrations with Notion, Linear, Confluence, Jira, and n8n, letting the Lovable Agent pull context from those tools and build prototypes, with a Miro connector coming soon. The company also began rolling out a new Design view with themes, expanded visual editing, and AI-powered image generation, with availability expanding to all users by the end of the week.

### Vercel — Vercel collaborates with Google for Gemini 3 Pro Preview launch

- Company: Vercel (vercel.com)
- Announced: 2025-11-18T13:00:00+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/vercel-collaborates-with-google-for-gemini-3-pro-launch
- Record: https://forck.live/items/20789-vercel-collaborates-with-google-for-gemini-3-pro-preview-launch
- Subject: v0 / AI SDK / AI Gateway

Vercel has made Google's Gemini 3 Pro Preview model available through its AI Gateway and on v0.app, following several weeks of internal testing. The model shows a 17% increase in correctness over its predecessor on Vercel's Next.js evals and ties with openai/gpt-5-codex at a 42% success rate.

### Lovable — How Delivery Hero Green-Lights Features 66% Faster with Lovable

- Company: Lovable (lovable.dev)
- Announced: 2025-11-18T12:51:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://lovable.dev/blog/deliveryhero
- Record: https://forck.live/items/12121-how-delivery-hero-green-lights-features-66-faster-with-lovable
- Subject: Lovable / AI app builder

Delivery Hero's consumer-facing product team uses Lovable to build interactive prototypes for stakeholder alignment, reducing the time to green-light features from three weeks to one week. The team built a working prototype in one hour and achieved directional alignment with customers and internal teams in a single meeting.

### Lovable — How n8n accelerates product development with Lovable

- Company: Lovable (lovable.dev)
- Announced: 2025-11-18T11:50:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://lovable.dev/blog/n8n
- Record: https://forck.live/items/12122-how-n8n-accelerates-product-development-with-lovable
- Subject: Lovable / AI app builder

n8n's product management and design teams use Lovable for rapid prototyping, building internal tools, and aligning on the product roadmap. The team reports faster idea development and the ability to generate deployable code, reducing the need for separate high-fidelity versions.

### Lovable — How Sentry uses Lovable to build what can’t be bought

- Company: Lovable (lovable.dev)
- Announced: 2025-11-18T11:43:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://lovable.dev/blog/sentry
- Record: https://forck.live/items/12123-how-sentry-uses-lovable-to-build-what-can-t-be-bought
- Subject: Lovable / AI app builder

Sentry's CRO Chris De Vylder used Lovable to build an ROI calculator and a sales dashboard for his team, addressing limitations of legacy sales tools. The ROI calculator took about one hour to build and another to fine-tune, and the entire sales team has adopted it.

### Lovable — How Thinkific uses Lovable to prototype the future

- Company: Lovable (lovable.dev)
- Announced: 2025-11-18T11:27:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://lovable.dev/blog/thinkific
- Record: https://forck.live/items/12125-how-thinkific-uses-lovable-to-prototype-the-future
- Subject: Lovable / AI app builder

Thinkific's product team uses Lovable to turn static mockups into interactive prototypes, enabling faster usability testing and earlier private beta launches. The team reports saving roughly five days of design effort per cycle and achieving a 94% average task-completion success rate in usability tests. Lovable allowed Thinkific to begin its private beta approximately three months earlier than planned.

### OpenAI — Intuit and OpenAI join forces on new AI-powered experiences

- Company: OpenAI (openai.com)
- Announced: 2025-11-18T05:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/intuit-partnership
- Record: https://forck.live/items/380-intuit-and-openai-join-forces-on-new-ai-powered-experiences
- Subject: GPT / ChatGPT / API

OpenAI and Intuit have entered a multi-year partnership worth over $100 million to launch Intuit app experiences in ChatGPT and expand Intuit's use of OpenAI's frontier models for personalized financial tools.

### Inception — Mercury Diffusion LLM Now Available on Azure AI Foundry

- Company: Inception (inceptionlabs.ai)
- Announced: 2025-11-18
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.inceptionlabs.ai/blog/mercury-azure-foundry
- Record: https://forck.live/items/18526-mercury-diffusion-llm-now-available-on-azure-ai-foundry
- Subject: Mercury

Inception made its diffusion large language model Mercury available on Azure AI Foundry for enterprise developers. The model features a 128K token context window, native tool calling, and structured output support, with a software license cost of $0.78/hour plus separate compute charges.

### Genspark — Genspark: A Strategic Collaboration with Microsoft to Bring AI Agents to Billions of Knowledge Users Worldwide

- Company: Genspark (genspark.ai)
- Announced: 2025-11-18
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.genspark.ai/blog/genspark-microsoft-agent365
- Record: https://forck.live/items/18264-genspark-a-strategic-collaboration-with-microsoft-to-bring-ai-agents-to
- Subject: Genspark

### Factory AI — Palo Alto Networks and Factory: Secure AI-Powered Development at Scale

- Company: Factory AI (factory.com)
- Announced: 2025-11-18
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://factory.com/news/palo-alto-networks-partnership
- Record: https://forck.live/items/17866-palo-alto-networks-and-factory-secure-ai-powered-development-at-scale
- Subject: Droid

Factory integrated its agent-native development platform with Palo Alto Networks' Prisma AIRS to provide real-time inspection of prompts, model responses, and tool calls, aiming to secure AI-powered development workflows against prompt injection and other threats.

### Cua — Introducing the Cua CLI: Manage Cloud Sandboxes from Your Terminal

- Company: Cua (cua.ai)
- Announced: 2025-11-18
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cua.ai/blog/introducing-cua-cli
- Record: https://forck.live/items/16829-introducing-the-cua-cli-manage-cloud-sandboxes-from-your-terminal
- Subject: Cua Driver / Lume / Cua Bench / Cua Fleets

Cua launched a CLI that lets users create, manage, and connect to Linux, Windows, or macOS cloud sandboxes from the terminal, supporting both flat and grouped command styles, non-interactive mode for CI/CD, and integration with the company's Python and TypeScript SDKs.

### Amp — Gemini 3 Pro

- Company: Amp (ampcode.com)
- Announced: 2025-11-18
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/gemini-3
- Record: https://forck.live/items/7784-gemini-3-pro
- Subject: Amp

Amp switched the main model powering its smart agent mode from Claude to Gemini 3 Pro. The change resulted in a 17 percentage point improvement on Terminal-Bench 2.0. A fallback setting for Sonnet 4.5 was subsequently removed.

### Google DeepMind — WeatherNext 2: Our most advanced weather forecasting model

- Company: Google DeepMind (deepmind.google)
- Announced: 2025-11-17T15:09:23+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://deepmind.google/blog/weathernext-2-our-most-advanced-weather-forecasting-model/
- Record: https://forck.live/items/1277-weathernext-2-our-most-advanced-weather-forecasting-model
- Subject: Gemini

A new AI model provides more efficient, accurate, and higher-resolution global weather predictions.

### ElevenLabs — Introducing ElevenLabs Image & Video

- Company: ElevenLabs (elevenlabs.io)
- Announced: 2025-11-17T12:00:00+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://elevenlabs.io/blog/introducing-elevenlabs-image-and-video
- Record: https://forck.live/items/3376-introducing-elevenlabs-image-video
- Subject: Voice AI

ElevenLabs introduces a product named ElevenLabs Image & Video.

### OpenAI — OpenAI named Emerging Leader in Generative AI

- Company: OpenAI (openai.com)
- Announced: 2025-11-17T10:00:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/gartner-2025-emerging-leader
- Record: https://forck.live/items/381-openai-named-emerging-leader-in-generative-ai
- Subject: GPT / ChatGPT / API

OpenAI was recognized as an Emerging Leader in Gartner's 2025 Innovation Guide for Generative AI Model Providers.

### Cua — Cloud Windows Sandboxes GA + macOS Preview

- Company: Cua (cua.ai)
- Announced: 2025-11-17
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cua.ai/blog/cloud-windows-ga-macos-preview
- Record: https://forck.live/items/16830-cloud-windows-sandboxes-ga-macos-preview
- Subject: Cua Driver / Lume / Cua Bench / Cua Fleets

Cua made Cloud Windows Sandboxes generally available and opened early preview access for Cloud macOS Sandboxes. The Windows sandboxes provide a full Windows 11 desktop in a browser with hot-start under one second, persistent state, and three size tiers. macOS sandboxes are invite-only and provision bare-metal M1, M2, or M4 hosts on demand.

### xAI — Grok 4.1

- Company: xAI (x.ai)
- Announced: 2025-11-17
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://x.ai/news/grok-4-1
- Record: https://forck.live/items/4350-grok-4-1
- Subject: Grok

xAI introduces Grok 4.1, a new model with significant improvements in creative, emotional, and collaborative interactions, achieving top scores in LMArena Text Arena and demonstrated preference over previous models. Available on grok.com, X, and mobile apps.

### Replicate — Replicate is joining Cloudflare

- Company: Replicate (replicate.com)
- Announced: 2025-11-17
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replicate.com/blog/replicate-cloudflare
- Record: https://forck.live/items/2473-replicate-is-joining-cloudflare
- Subject: Platform

Replicate announces that it is joining Cloudflare. It will continue as a distinct brand, the API will not change, and existing models will keep working. The integration aims to make Replicate faster and better integrated with Cloudflare's Developer Platform.

### Hugging Face — Easily Build and Share ROCm Kernels with Hugging Face

- Company: Hugging Face (huggingface.co)
- Announced: 2025-11-17
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/build-rocm-kernels
- Record: https://forck.live/items/1586-easily-build-and-share-rocm-kernels-with-hugging-face
- Subject: Platform

Hugging Face published a guide on building, testing, and sharing ROCm-compatible kernels using its kernels library, focusing on the kernel-builder tool and the RadeonFlow GEMM kernel as an example.