Copy/Paste is here + 14 new models

You're receiving this email because you registered to one of our workshops. You can unsubscribe at the bottom of each email at any time.

This week, MindStudio released copy/paste + 14 new models from Meta, Mistral, Google, Stable Diffusion and Eleven Labs. Copy/paste has been on our feature request board for a long time, and we’re sure power users will love it. MindStudio is live on G2 now, and we’d deeply appreciate if you could take a moment to help us with a review.

In the last couple of weeks, we got new models from Mistral, new and cheaper GPT-4o and Gemini 1.5 Flash snapshots, GPT-4o long output, Google released Gemini 1.5 Pro Experimental topping the chatbot arena, there was a major scare of the AI hype bursting, and Canva acquired Leonardo.

Continue reading to learn more!

As a reminder, we’re now welcoming partners that want to build AIs for their clients. Sign up for extra support, training resources, and more here.

🗞️ Industry news

1. GPT-4o updates: price cuts and longer outputs
   
   

   

   Posted by OpenAI on Twitter.

   
   
   OpenAI is on an open pricing battle with Google. 2 years ago, GPT-4 was prohibitively expensive for most use cases. Now, OpenAI offers GPT-4o mini at less than $0.10/1Mtks output and decided to heavily discount their top-tier GPT-4o model as well.
   
   When models get cheaper, they get cheaper in MindStudio as well. We’re working to include the new snapshot and update our pricing for GPT-4o in MindStudio.
   
   Other than the price cut, the new snapshot also include better JSON handling, an alpha of a 64k tokens output (content people will love this!), and lots of new security assurances.
   
   Let’s be blunt for a sec here. OpenAI… kinda lost its edge. All competitors now reached GPT-4 & GPT-4o level, with a couple actually surpassing them. Employees at OpenAI already said GPT-5 is NOT coming in Autumn, and Mira Murati (CTO @ OpenAI) hinted at a late 2025 release date.
   
   We might need to get used to a world where OpenAI is not the innovator anymore. Or maybe they’re cooking something awesome and the strawberry pic from Sam Altman is an indication.
   
   

2. Google is the new chatbot arena leader
   
   

   

   
   
   Since GPT-4o mini surpassed Claude 3.5 Sonnet in the arena, I kind of lost interest in their ratings and refer to Artificial Analysis instead. Claude 3.5 Sonnet is my favorite model right now, and its quality is most definitely higher than GPT-4o mini. People might just be biased towards GPT-like answers.
   
   Regardless, this week we got a new leader: GPT-1.5 Pro Experimental. With an ELO around 1300, it outperforms GPT-4o and Claude 3.5 Sonnet, the current market leaders, and the previous Gemini 1.5.
   
   Together with Gemini 1.5 at 2 million context window, the release shows Google is finally starting to get serious about shipping actual useful models at a good pace. They even discounted Gemini 1.5 Flash massively, and basically matched the cost GPT-4o mini while offering a 1 million context window (vs the 128k context window of 4o mini).
   
   That being said, we’re talking about marginal improvements here. You won’t have an “awe” moment using Gemini 1.5 Pro Experimental compared to the other top tier models. For non-enthusiasts, it doesn’t look like the new model will be all that different from the current.
   
   The current Gemini 1.5 Pro model is already a top model, and I highly suggest you give it a go. Google radically improved its answers, and even if we probably aren’t going to see Gemini 1.5 Ultra anytime soon, the Pro family is enough for nearly all workflows, and performs particularly well for creative workflows.
   
   

3. Is the AI hype fading? Was it all just a big bubble?
   
   

   

   last month performance of the Tech stock market, visualized by FinViz

   
   This week, the stock market crashed after Japan reported dramatically low figures, Warren Buffet decreased his stake in Apple, and the NASDAQ dropped over 10%. Most of the crush was associated with the top tech companies, with all big names losing significant market share.
   
   The crash led to the media and social networks claiming the “AI bubble” is bursting. With VC funds stating AI isn’t even close to being profitable and investments slowing down, it’s a fair question to ask.
   
   So, is AI just a fad?
   
   In my opinion, not really. It’s just not a Terminator waiting to kill us all. Expectations for AI got so hugely blown out of proportion people are now bashing it for not being a sort-of-God yet. Do you remember how many newspapers and creators were talking about AI killing us all two years ago?
   
   People always overestimate the impact of technology in the short term and dramatically underestimate the change it will create in years.
   
   The same complaints we see now for AI, mostly related to how tech companies handle data, how they package their products, the EU writing thousands of pages to regulate something that doesn’t exist yet, etc. is something we’ve seen already. It’s nothing new.

   
   

   

   CNET about Google in Sep 2003

   
   Newspapers were opposed to Google scraping their content to index it. People were against Featured Snippets in search. We had Google cars roaming around the entire world taking pictures of houses, people, and events. The concept of saving your credit card in a web browser or banking from your sofa was considered crazy. And don’t get me started on a smartwatch that tracks your location 24/7.
   
   Generally speaking, writing against a technology brings you more clicks than writing in favor of it. Additionally, most new tech fails, so you’d be right more often than not.
   
   But useful technology evolves, usually in leaps. And, slowly but surely, useful tech innovations find their way into the mainstream and create more opportunities for everyone that embraces it.
   
   Many are even comparing AI to crypto. I worked in crypto before, and I can confidently say the impact is completely different. Crypto and the blockchain didn’t “fail”, they failed for the masses. They tried to solve a problem (banking and privacy) Western Countries don’t really have anymore. It’s still used in developing countries or war-torn nations, but there’s very little reason for an American or a European to use a crypto wallet versus their bank account.
   
   With AI, it’s very different. AI can help quite literally every person do more, better. There are friction points like hallucinations, cost (now less of a problem), and ethical/environmental concerns that stop long-term adoption, but I’m positive we’ll solve these because the positive outcomes are too big to ignore.

🔥 Product Updates

When starting a new MindStudio project or working on multiple projects at once, you can now copy over components from one application to another, including:

• Automation blocks

• Prompts

• Configurations

• Model settings

• User inputs

• Functions

• Entire workflows

When you copy a block, you’re effectively copying a JSON structure that MindStudio translates into workable templates. This means you can easily share prompt libraries with others in your team or online, create a prompt library as a chrome extension to copy-paste blocks and functions, or sell access to a set of templates in MindStudio.

I asked Claude 3.5 Sonnet to create a Chrome Extension to store my favorite MindStudio blocks in JSON and let me copy and paste them in the builder. It did it in three shots, with a Tailwind UI. Give it a go if you use extensions frequently and you find yourself copy/pasting the same blocks often.

We also shipped 14 new models in the platform: 

• Llama 3.1 8b Instruct

• Llama 3.1 70b Instruct

• Llama 3.1 405b Instruct (best open source model!)

• Mistral Nemo

• Mistral Codestral

• Mistral Codestral Mamba

• Mistral Large 2

• Gemini Pro Vision

• Gemini 1.5 Pro Vision

• Gemini 1.5 Flash Vision

• Stable Image Core

• Stable Diffusion 3

• Stable Image Ultra

• Elevenlabs TTS

The new Vision models give you an alternative if you don’t get satisfying results from OpenAI’s GPT-4o and GPT-4 turbo, and the Eleven Labs voices are typically higher-quality than OpenAI’s TTS - give them a go!

💡 Tip of The Week

Ever wondered what’s a “chunk” and how it relates to RAG? That’s a terminology we use to fetch specific portions of a document using RAG: retrieval augmented generation.

In MindStudio, that’s typically a Query Data Source block. This block will fetch data from a data source and return 1 to 5 chunks in a variable. Then, you can use that variable as context to ground your AI in reality or perform additional data cleaning operations.

A chunk in MindStudio is approximately 500 words, so the maximum output is around 2500 words. RAG will try to fetch the chunks that most closely resemble the query.

For example, if you have an encyclopedia as data source and your query is “cat”, RAG will try to find the chunks of the encyclopedia that contain the word cat. It will run the query X times (up to 5) and save the result in a variable you define in the Query Data Source block.

Then, you can use that variable in a Generate Text block, or even in the chat terminator, to let the AI know more about the query. In this example, cats.

You can query multiple data sources in a workflow for different queries, but you should avoid querying data sources that only contain a couple of chunks.

Let’s assume your encyclopedia is split into 2 chunks of 500 words each, for a total of 1000. When you query it, you ask for 3 chunks. That’s simply not possible, given that the data source is only split into 2.

And that’s not all. If a file is so small it fits within 1 to 3 chunks, you’re probably better off adding that as context within the prompt rather than running a semantically related search (RAG) to find the appropriate pieces. In most cases, RAG will simply retrieve the whole document anyway.

If you want to learn more about RAG, you can watch our short explainer video on YouTube (10m) or watch a full webinar on the topic (50m).

🤝 Community Events

If you want to hangout with our team, we usually host a Discord event every Friday @ 3PM Eastern. Join our Discord channel to keep up to date with the hangouts - our entire team is active there.

You can register for upcoming events on our brand new events page here.

Our new webinar series is up on there as well, with the following on-demand webinars:

• MindStudio for Partners: Level Up Your Agency With AI Workflows

• Customize AI Applications With Your Data

Plus, we have new weekly and bi-weekly events:

• How To Build Your First AI Assistant (the best webinar for beginners!)

• Which Models to Choose for Your AI

• First Steps to Bring AI Into Your Workflow

Thank you for being an invaluable member of our community, it’s always great to see many of you join multiple workshops 🔥

If you’re interested in any topic in particular, feel free to reply and I’ll do my best to include it in the next releases. We’re going to update all of these soon.

🌯 That’s a wrap!

Stay tuned to learn more about what’s next and get tips & tricks for your MindStudio build.

You saw it here first,

Giorgio Barilla
MindStudio Developer & Project Manager @ MindStudio