Introduction

You’re there, casually scrolling your timeline as a subtle vibration from your phone catches your attention. Another AI LLM is reaching the market and is the new best toy in town that everyone should check.

The first time that happened, you were probably actually intrigued and maybe even went to check it out, but, by the fourth and even fifth time, checking new LLMs and AIs started to become a little tiring.

Want to check which are the best LLMs for each case (API, personal use/prompting, etc), ranked, without having to spend hours searching and comparing each one? This article is for you, then.

Criteria for Ranking

The ranking of language models in this leaderboard is based on a list of criteria made by me, designed to evaluate their overall performance and applicability. These criteria include:

• Overview: A small resume about who is the creator of the LLM (and maybe even some polemics you should learn about)
• Strengths: Where and in which cases the LLM excels at
• Cost: The model's capacity to handle increasing amounts of data and user interactions along with how much would that cost you. Generally, this uses tokens for measurement. Not sure what a token is and how it’s calculated? You can consider it equal to ~4 characters of text for common English.
• Ease of Integration/Community Support: How easily the model can be integrated into existing systems and workflows and the availability of resources, documentation, and community engagement to assist us with implementation and usage.

LLM Setup - The quick and easy route

HOLD FAST! Before we continue, if you’d prefer to learn by doing and using those powerful LLMs we have something exciting for you!

Join our waitlist for early access to Latitude’s new platform. It's designed to help LLM developers streamline their work and bring ideas to life with confidence, helping a ton in the measurement process I mentioned above.

We're honestly truly excited about this platform and can't wait to see what you create with it!

Don't miss out! Click the link below to join our waitlist 🚀

⭐️ Enter the Waitlist! ⭐️

Top Language Models

Now, let’s jump right into the LLMs that are most famous today. Of course, I wouldn’t be able to actually review every single one (there are literally thousands of models out there) but, here we’ll check: GPT, Gemini, LLaMA, Claude, and Copilot.

GPT

Overview: GPT (Generative Pre-trained Transformer) is one of the most well-known language models developed by OpenAI. It has a ton of models: 3.5-turbo, 4, 4o (the most recent), and more. In terms of polemics, it’s involved in kind a few of them (e.g. ChatGPT's voice closely resembles Scarlett Johnasson's, says lab analysis : NPR), so, you can also take that into consideration.

Strengths: Exceptional accuracy and versatility in generating human-like text. It has some great API documentation and integration also, while still maintaining a good interface for human interaction.

Prices: Varies depending on usage and subscription plans, the better trained the model, the more expensive it becomes, ranging from $0.5/1M tokens to $5/1M tokens on the most recent model. You can check them all here: Pricing | OpenAI

Looking for image models? It also has, with prices ranging from $0.016/image to $0.040/image. If you’re interested in using it for prompting and normal use instead of integration, it has both a free plan and a Plus plan with different models, image generation, and a price of $20/month.

Gemini

Overview: Gemini is a robust language model known for its efficiency and scalability. It’s made by Google and recently had some really awesome announcements at I/O (a conference made by Google to showcase new stuff). It’s also mentioned in a few polemics (e.g Google pauses Gemini's AI image generation after diversity controversies) that you can take into consideration.

Strengths: Amazing context window (this basically means it can use more data as the font of a single response), great language proficiency too!

Prices: For APIs, pricing based on usage too, having a free tier with low requests number and $0.35 / million tokens, later on. For personal use, Gemini Code Assist, for example, costs $19 per user, per month.

LLaMA

Overview: LLaMA (Large Language Model Meta AI) is designed to be less resource-intensive than other models, it has a different license scheme, which means that this resembles “something open source” and as the other ones, also has a few polemics in its name (e.g. Meta Stops Disclosing the Data It Uses to Train AI Models Like Llama 2 - Business Insider).

Strengths: They try to follow guidelines for responsible AI creation and innovation. It also has different models specific for different scenarios like Code LLaMA, for code generation, for example. It’s great for research and English-focused applications.

Prices: Basically free while using Meta AI or downloading and running it locally/self-host. You can also check out services for using as API, that for example can generate API keys for using LLaMA.

Claude

Overview: Claude is a versatile language model that excels in understanding and generating natural language. It’s a really interesting alternative to GPT, honestly — and, for the first time, I couldn’t find some polemics including it.

Strengths: Seems to perform better than GPT mostly regarding speed and prices, but tries to reach the same strengths, like conversational AI, etc.

Prices: For personal use, it starts out free and grows up to $20 for pro or $30 for teams. For APIs, costs $3/M tokens for input and $15 for output on their most intelligent model to date.

Copilot

Overview: Microsoft Copilot (here, I’ll evaluate both Copilot for Microsoft 365 and GitHub Copilot) and, although this one is not necessarily an LLM, it’s been publicly known that Microsoft is developing an in-house AI LLM — that is probably going to be connected to these products in the future, and, since Windows is used by a ton of people, I’ll also leave my review of them here. And let’s not forget some polemics here too (e.g. Controversial Microsoft AI screenshot feature delayed over security concerns | Science & Tech News | Sky News)

Strengths: Aims to be the best in aiding people on the daily life, from help with emails, to Excel, to code with GitHub Copilot’s suggestions and autocomplete. The strong side here is obviously if you/your company use Microsoft’s services. Outlook, Excel, Word, Teams, and all other products can benefit from here.

Prices: You'll actually find a really cool free tier here for consumption (prompting and images , but not code). But, if you want more of that uniqueness of integrating all Microsoft's products with AI, it's a subscription-based model with various tiers. You can check them all here: GitHub Copilot · Your AI pair programmer and here. Honestly, they both seem kind of expensive with GitHub Copilot seeming a little bit less, but, in company terms, they’re probably not that expensive.

Conclusion

Awesome! So, I hope that throughout this article I provided an in-depth look at the top language models, highlighting their strengths, pricing, polemics, and use cases. With Latitude's tools and insights, developers have the resources they need to choose the best LLM for their projects and optimize their performance, so, make sure to join the waitlist!

We encourage you to explore these models further and take advantage of Latitude's platform to enhance your development process. Don't forget to share your thoughts and experiences in the comments, and join the Latitude community to stay updated on the latest advancements in AI and LLMs.