- Horizon AI
- Posts
- The Best Open-Source AI Model is Here 🔥
The Best Open-Source AI Model is Here 🔥
+ Convert a script into a speech in seconds
In partnership with
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/2813bdf6-0c67-4e7c-b1bc-1bf2e60c41ae/Logo_black_banner_v3.png?t=1720030963)
Welcome to another edition of Horizon AI,
In April, Meta teased an open-source AI model aiming to match the performance of top private models, and today it has arrived.
Let’s get right into it!
Read Time: 4.5 min
Here's what's new today in the Horizon AI
Meta's Most Powerful Open-Source Model Yet
Kling AI Video is Now Available Worldwide 🌐
AI Tutorial: Convert a script into a speech in seconds
AI Tools to check out and AI Job board
The Latest in AI and Tech 💡
AI Findings/Resources
AI News
META
Meta's Most Powerful Open-Source Model Yet
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/22a2f319-1358-4841-85ae-089af634524d/ca072de4_67b7_406d_b8cd_004b128ed7c0.jpg?t=1721830872)
Meta has announced the release of Llama 3.1, the largest open-source AI model to date, which the company claims outperforms both GPT-4o and Anthropic’s Claude 3.5 Sonnet on several benchmarks.
Details:
Llama 3.1 boasts 405 billion parameters and was trained with over 16,000 Nvidia H100 GPUs.
It delivers stronger reasoning abilities, a larger 128K context window, and support for new languages, including French, German, Hindi, Italian, and Spanish.
The model will be available through WhatsApp and the Meta AI website in the US, with availability on Instagram and Facebook to follow.
Model weights are being released to allow companies to train and fine-tune Llama 3.1 on custom data.
This release is a significant step in Meta’s AI strategy, with CEO Mark Zuckerberg predicting that Meta AI will become the most widely used assistant by the end of the year, surpassing ChatGPT.
TOGETHER WITH WEBSTREET
Cash is king 👑
Not long ago, investors were chasing rocketship startups in hopes of a huge payday.
Boy how times have changed.
Those overvalued, buzzy startups just aren't as attractive anymore. You know what's cool? Undervalued companies that earn real profit.
WebStreet finds these companies, acquires them, improves them, and sells them. (And lets you invest)
They buy overlooked businesses already earning money
They improve operations, so they make even more money
They bundle everything into a single, powerful fund that you can invest in. It's like micro private equity on autopilot.
Unlike a lot of startups, WebStreet is doing great.
Are you an accredited investor looking for passive income and want to invest in businesses providing real value? Join WebStreet’s Waitlist Here.
KUAISHOU
Kling AI Video is Now Available Worldwide 🌐
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/7e7eadc6-5d74-43ab-9938-10ff94b38e94/ezgif-7-15ffc37609.gif?t=1721830292)
Is there an AI better than OpenAI’s Sora? Reports suggest that China-based Kling might be the best video-producing AI in the world.
Previously available only in China, KlingAI has now launched version 1.0 of its AI video platform worldwide. It no longer requires a Chinese phone number.
Details:
Kling features both Text-to-Video and Image-to-Video capabilities.
Currently supports high-performance options with future updates expected to allow for higher quality.
Only a 5-second video option is currently available.
Registration requires just an email address.
Daily logins will grant users 66 free credits for video creations; 10 credits are needed to create a video.
You can try out Kling AI here.
AI Tutorial
Convert a script into a speech in seconds
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/4c0c419a-802b-4cc7-a7a9-25b867029aee/image.png?t=1721826619)
We'll be using Parler-TTS, a free text-to-speech model that can generate high-quality, natural-sounding speech in the style of a given speaker.
The Parler-TTS model is available on the Hugging Face website
Write your text prompt and the description of the voice you'd like
You can find examples if you scroll down
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/f6ba5550-1f69-412e-b160-4566fddae679/image.png?t=1714746836)
You can also find tips on the same page to ensure you get exactly what you want
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/a6926a94-0688-482b-91b8-12a58fdbfb22/image.png?t=1714747179)
Once you are done just click on the "Generate Audio" button and wait for the audio to be generated.
Listen to the generated audio, and if necessary, make changes to the prompt or voice descriptions. You can then download or share it
AI Tools to check out
🎧 Wondercraft: An AI powered audio studio.
🚀 MindPal: Build AI agents for your tasks, with your data.
🗣️ Speax: Dub your videos into different languages in few clicks.
✅ KrutAI: AI co-pilot for e-commerce brands.
Whenever you are ready, here's how we can help you:
Book an Audit call with our AI Automation agency: Our clients saved 40+ hours in their first month. Book your call now, Limited spots available. (Your business must be making a minimum of $5K/month)
AI Job Board
💻 Software Engineer - Crunchyroll
📈 Technical Product Manager - The New York Times
👨💻 Software Engineer - Shield AI
💻 Senior Data Analyst - Aircall
AI Findings/Resources
🔎 Gen AI coming for remote workers first?
🤖 Tech experts detail 12 ways pop culture misrepresents AI’s abilities.
✨ Build a full web app with Claude with 2 screenshots.
The latest in AI and Tech
Adobe is introducing new generative AI tools to Illustrator and Photoshop to enhance creative workflows.
A group of five U.S. senators has sent a letter to OpenAI CEO Sam Altman, requesting detailed information on how the company plans to fulfill its safety and security commitments.
Antitrust authorities from the FTC, DOJ, UK, and EU have issued a joint statement to ensure fair competition in the AI industry. They highlight concerns that AI chipmakers might exploit bottlenecks, potentially gaining disproportionate influence over the future development of AI tools.
A study by the Data Provenance Initiative reveals a rapid increase in web domains blocking AI models from accessing training data. The percentage of blocked tokens has surged from 1% to 5-7% in one year. This trend could hinder AI development, forcing reliance on less reliable data.
AI Showcase
🖼 AI art of the day
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/effea326-2c0f-49d4-9cf4-7d5c8149851e/0_2-midjourneyy.jpeg?t=1721832233)
Source: furious0450 on Midjourney
A red and pink colorful fantasy movie poster of an armored samurai warrior in the style of Yoji Shinkawa, in pink and red, holding his sword with both hands, standing on one leg while leaning to the left side, full body shot, detailed texture, concept art in the style of Alex Ross. --ar 16:9 --stylize 750
📹 AI video of the day
Get ready to laugh! @Pizza_Later made a hilarious and super cute AI short film that’s like The Office but with puppets that go “poof”—quite literally…
This fun-filled film has awesome consistency and will have you laughing from start to finish.
Who knew puppets could be this… x.com/i/web/status/1…— Curious Refuge (@CuriousRefuge)
9:11 PM • Jul 16, 2024
That’s a wrap!
We'd love to hear your thoughts on today's email!Your feedback helps us improve our content |
Not subscribed yet? Sign up here and send it to a colleague or friend!
See you in our next edition!
Gina 👩🏻💻