• FavTutor
  • Posts
  • Midjourney opens website to all users

Midjourney opens website to all users

Including the latest AI news of the week

Hello, AI Enthusiasts!

Welcome to FavTutor’s AI Recap! We’ve gathered all the latest and important AI developments from the past week in one place, just for you.

In Today’s Newsletter: 😀 

  • Midjourney's AI Image Generator is now Free*

  • Coder Solves 633 Leetcode Problems in 24 hours using Claude AI

  • Fine-tuning is now available for GPT-4o

  • Prompt Caching in AI Can Save Up To 90% on API Costs

  • Microsoft’s Phi-3.5 models are better than many LLMs

Midjourney
🖼️ Midjourney's AI Image Generator is now Free*

Midjourney is one of the most used AI Image Generator today, and recently they launched a website version to use it. Now, Midjourney CEO David Holz announced via a Discord message that they are offering free trials for everyone to try it.

Insights for you:

  • New users on the Midjourney website can create about 25 images for free.

  • This will lure more people into trying their services without any payment.

  • Users can also register using their Google account along with the Discord account.

Together With Guidde
Your team + AI-generated video documentation

Is your team explaining the same thing over and over again to your colleagues, clients and new hires?

Imagine this: your team delegates that work to AI with Guidde – the Gen AI app that helps your team explain the most complex tasks in seconds. It does this by:

  • Automatically creating a storyline with highlights

  • Adding and translating text to voice in over 40+ languages

  • Including easy share options as a link, pdf, mp4, html…

  • Integrating with Zendesk, Notion, Slack, Salesforce and more.

Simply click capture on the browser extension and the AI will automatically generate step-by-step video guides, complete with visuals, voiceover, and call to actions.

The best part? You can try Guidde for free

Coding
🤯 Coder Solves 633 Leetcode Problems in 24 hours using Claude AI

Leetcode is considered to be one of the best platforms for solving DSA-related problems, that companies use to test the skills of candidates for entry-level jobs. However, a person used Claude AI and Python to solve these problems with an automated system.

Insights for you:

  • A Reddit user named “TimS2024” has automated Leetcode using Claude 3.5 Sonnet API and a small Python script.

  • He solved 633 problems in 24 hours with a success rate of 86%.

  • This also sparked the debate on whether AI will replace Software Engineers or not.

OpenAI
⚡️ Fine-tuning is now available for GPT-4o

Developers can now, fine-tune GPT-4o to improve the model’s structure and domain-specific instructions for their AI applications. Developers can already produce strong results for their applications with as little as a few dozen examples in their training data set.

Insights for you:

  • OpenAI has launched fine-tuning for the GPT-4o model.

  • They are also offering 1M training tokens per day for free for every organization through September 23.

  • Fine-tuning can have a large impact on model performance for coding to creative writing.

Anthropic
💰️ Prompt Caching in AI Can Save Up To 90% on API Costs

Prompt caching means storing prompts at a temporary location for easy access. When you type a prompt on LLMs like ChatGPT or Claude, they must process the request and generate a response by doing many complex computations.

Insights for you:

  • Anthropic introduces Prompt Caching; many are calling it a game-changer.

  • It helps reduce costs by up to 90% and latency by up to 85% for long prompts.

  • Claude and Gemini offer prompt caching features, but not OpenAI.

Microsoft
💪 Microsoft’s Phi-3.5 models are better than many LLMs

Microsoft released Phi-3.5-MoE which seems to have competitive reasoning capabilities, beating Google, OpenAI, Meta or Mistral. Phi-3.5-MoE outperformed Google’s Gemma 2 9B, and did came close to OpenAI’s GPT-4o mini.

Insights for you:

  • The 3 new Phi 3.5 models are Phi-3.5-mini-instruct, Phi-3.5-MoE-instruct, and Phi-3.5-vision-instruct.

  • They come in in 3.8 billion, 4.15 billion, and 41.9 billion parameter versions.

  • These models are available to download and fine-tune on Hugging Face under a Microsoft-branded MIT License.