Mira Murati's AI Startup Lands Massive $2B Seed 💰

Welcome to another edition of Horizon AI,

After an abrupt exit from OpenAI last September, Mira Murati’s six-month-old company is already valued at $12 billion and is preparing to launch its first product.

Let’s get right to it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Mira Murati’s AI Startup Secures $2 Billion in Funding

• Mistral Launches Open-Source AI Audio Model

• AI Tutorial: How to create multilingual subtitles using Gemini

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

THINKING MACHINES LAB

Mira Murati’s AI Startup Secures $2 Billion in Funding

Mira Murati, the former chief technology officer of OpenAI, just announced that her startup is preparing to unveil its first product “in the next couple of months” after raising $2 billion in fresh capital.

Details:

• The deal values the startup at $12 billion, with Andreessen Horowitz’s a16z leading the funding round, joined by Nvidia, Accel, ServiceNow, Cisco, AMD, and Jane Street.

• The company is focused on developing multimodal AI that interacts with people in a “natural way” — through conversation, vision, and “the messy way people collaborate.”

• Murati added that the upcoming product will include a significant open-source component and will be “useful for researchers and startups developing custom models.”

• The company also plans to share its “best science” to help the research community better understand frontier AI systems.

Murati left OpenAI in September and launched Thinking Machines in February. While few details have been shared publicly, the startup has clearly impressed investors with its vision and strong team, which includes former employees from OpenAI, Google, and Meta.

Source

MISTRAL

Mistral Launches Open-Source AI Audio Model

French AI startup Mistral has released Voxtral, its first open-source family of audio models designed for businesses.

Details:

• Voxtral is positioned as a low-cost alternative to closed systems, with transcription services priced at less than half the cost of competitors.

• The model can transcribe up to 30 minutes of audio and understand up to 40 minutes, enabling features like summaries, voice commands, and real-time actions.

• Voxtral supports multiple languages, including English, Spanish, French, Hindi, and more, making it versatile for global use.

• Two variants available:

  • Voxtral Small (24 billion parameters) for large-scale deployments.

  • Voxtral Mini (3 billion parameters) for local and edge use, with a stripped-down version optimized for transcription-only tasks.

Users can test Voxtral for free by downloading it on Hugging Face to run locally, or by using it in Mistral’s chatbot, Le Chat. API integration starts at $0.001 per minute.

Source

AI Tutorial

How to create multilingual subtitles using Gemini

1. Go to the Google AI Studio platform and sign in.

2. Choose 'Gemini 2.5 Pro' as your model.

3. Upload the video you want to work with and use this prompt:

"Analyze this video and provide a precise transcript with timestamps (HH:MM:SS) in [specify language]."

4. The tool will process the video and provide subtitles in your chosen language.

*Note: Gemini may sometimes fail to provide properly timestamped transcripts on the first request. If that happens, simply repeat the prompt, and it will provide the correct output.

AI Tools to check out

📷 Filtrix: Instantly turn your photos into stunning art with AI.

🔊 VoiSpark: Transform text into natural speech, modify voices, and create unique vocal identities.

👨‍💻 CodeFriends: Learn coding and AI, simply.

💥 Tidio: Free AI assistant for your helpdesk.

💼 Mysite.ai: AI website builder for small businesses.

AI Findings/Resources

👀 Elon Musk’s xAI just posted the nerdiest job opening of all time

📝 10 use cases for AI agents: IoT, RAG, & disaster response explained

👉 4 ways businesses should use AI chatbots

🤔 A former OpenAI engineer describes what it’s really like to work there

💬 Google’s generative video model Veo 3 has a subtitles problem

The latest in AI and Tech

Meta continues AI hiring spree, poaches two top OpenAI researchers

Researchers Jason Wei and Hyung Won Chung will be joining Meta’s superintelligence lab, the latest in a series of high-profile OpenAI researchers and developers to be hired by the social media giant. Both had previously left Google’s Brain research group to join OpenAI in 2023.

Google says ‘Big Sleep’ AI agent found bug hackers planned to use

The AI agent, introduced last year to identify security vulnerabilities in both Google products and open-source projects, used information from Google Threat Intelligence to discover the issue before it could be exploited, according to the company.

Nvidia says it will resume H20 AI chip sales to China “soon”

The company is filing applications to restart H20 AI chip sales to China, saying the U.S. government has assured it that licenses will be granted and that deliveries could begin soon.

President Trump announces $92B investment in energy and AI

Mr. Trump and several leaders from the country’s largest technology and energy companies announced more than $92 billion in private sector investment in Pennsylvania — including data centers, energy projects, power infrastructure, and the personnel needed to fuel the nation’s AI ambitions.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻