Stop re-building chat clients, I already have one!
Ideally they would just run their own chatbot on different existing chat platforms that you could verify your API key with, but with my project you can at least run that chatbot yourself.
[0] - https://github.com/arcuru/chaz
[1] - https://jackson.dev/post/chaz/ (Blog Post)
Interrupting output (or streaming responses in general) won't work in a non-custom chat client, but I don't tend to use that anyways.
Disclaimer: I’m a small contributor to that repo
I also can't tell if matrix-chatgpt-bot supports sending images, but Chaz does.
Chaz supports switching the model used at any time as well, including in the middle of the chat. So you can setup many separate models and switch between them as necessary for different rooms.
Locally-run LLMs are very cool, but I think we have some way to go for smartphone performance to make datacenter-run (or beefy personal workstation for privacy!) ones obsolete.
Until then, I also really like the idea of making chatbots accessible via instant messengers. I find myself using the one in WhatsApp quite a lot these days, just because it's conveniently available on multiple platforms (with my history synchronized) and I can tag it in when chatting with friends. It also works on flights with zero-rated messaging.
Claude’s context window for Opus is remarkable and I’d say that GPT4 is vastly inferior as a result.
The second Anthropic ship Claude with a sandbox that can use python (or equiv) to take care of tasks, I’d happily end my ChatGPT subscription.
GPT4 has been here for over a year and it shows. But until other assistants can manipulate Excel data as easily… I’m stuck paying for it.
Claude is amazing for summarisation, copywriting and reasoned thinking. I love it.
This was originally GPT-4, Claude 3 Opus, and Gemini Advanced. I recently added Meta AI when they launched.
Right now I've sent 486 queries through the first three systems.
The clearest pattern to emerge is that Gemini is terrible, not on par with the other two. There hasn't been a single query that it was the only model who did well. Around 1/4 of the time it gives a clearly inferior answer to the others.
But between GPT-4 and Claude it's less clear. 31 of the 486 queries Claude provided a significantly better answer than the other two but 20 times GPT-4 provided the significantly better answer.
I do think that Claude is a slightly better model but right now it's not a clear enough advantage that I'd recommend it generally. I will say you can probably cancel you Gemini subscription if you're using it though.
For everyday non-dev users, Msty for MacOS lets you run split chats and compare each model's answers.
Maybe I should give it another try.
I also frankly will not extensively use a model that I can't turn off data storage for.
Only downside I’ve found is it doesn’t save chat history for privacy reasons, but you can download them.
Not complaining, just curious.
Edit: top iOS search result is imposter app asking for sub. Be careful
I’m assuming it uses whisper v3 plus gpt for post processing the whisper transcript.
I hope that with iOS 18 siri’s transcription is as good.
ChatGPT’s transcription is good enough that I can trust that even for a long 2 minute input, it’ll be 99-100% right. Whereas Siri I can’t trust that even for a single sentence.
Best ones to play with:
Midjourney
Suno AI (music)
Claude
ChatGPT
Perplexity (a better version of “ChatGPT with bing”)
Then some more that others seem to like but I haven’t used / haven’t gotten into: Character AI and Pika Labs
Is this supposed to be higher in the list or is that expected?
They should at least train it with info on what the product actually is, can’t really get information on its current capabilities.
