Budget-Friendly AI APIs - A Must-Watch Guide!
Вставка
- Опубліковано 30 тра 2024
- Discover the most affordable AI models to use via an API. In this video I look at the top AI models from OpenAI, Mistral, Anthropic, etc to see which ones offer the biggest bang for your buck. I also dig into some of the billing terminology like token, input prices vs output prices, etc.
---
Let Me Explain T-shirt: teespring.com/gary-explains-l...
Twitter: / garyexplains
Instagram: / garyexplains
#garyexplains - Наука та технологія
Such a useful overview. Thanks! I hadn’t realized how great the spread is.
Haiku offers decent value given the size of the context window
Where is gemini?
Gemini (pro) even has a free tier apparently.
Gemini's api is currently not available in Europe, but for cloud deployments that often doesn't matter.
Other than that, this was a great summary.
if you are targeting Europeand market and handling customer sensitive data like ANY e-commerce website for example, then GDPR applies and your stack must be deployed in Europe servers
I’d like to learn to use an LLM and maybe Claude Haiku could work? Say, if you had a thousand books and these were all based around a philosophical school like ‘Stoicism’ and then the output would be narrated by a dude like Seneca, that would be so cool.
Please suggest a free vscode copilot alternative!
Codeium maybe?
@@GaryExplains I will try!! Thanks!!
Vs code is free to use, maybe you meant open source?
@jaydeep-p I think you missed the word "copilot" when you read the original post. They are looking for a coding assistant to use in vs code.
AWS CodeWhisperer
This reminds me how underrated LLaMA is 😅
I agree with the sentiment, but it is limited. I asked LLaMa 2 the towel question and got: If it takes 3 hours for 3 towels to dry, it will take 9 hours for 9 towels to dry. This is because the time it takes to dry a towel does not change if you increase the number of towels being dried at the same time.
It created code that compiles for the question, "Write C code that can parse and evaluate a mathematical expression like (4+5)*2+1" but the code gives the wrong answer.
@@GaryExplains Got you. It is limited with it's reasoning and assertions, if I understand you right. But I found it to be pretty decent as a conversational AI in my limited interaction with it. Thanks for the response
And one needs decent system to run LLM's local. Beefy GPU with lots of VRAM. Good CPU, the more MEM the better.
No everyone has this gear.
I was testing LlaMa 2 via the Brave browser, it is a cloud instance of the model.
ChatGPT 3.5