Yes please Sir - add Deepseek. Thanks very much in advance.
I came on the community specifically to request this feature. I double checked to see if it was already added but you’re on top of things.
Every time I use Deepseek the last 5 days all I could think is I want these results on Pickaxe. This DeepSeek is getting way better results for me than the OpenAI models. I’m very impressed and excited to offer this to my clients as well.
Yes! Deepseek please, if you were to add that what would be a rough estimate on completion?
it’s no brainer I guess, cheaper and better. it offers more values to user. For wrapper to survive the only thing we can do is to integrate ask many models as possible. This cant be done by openai alone or deepseek alone. The more models there are competing each other the higher the likelihood for small players like us to survive
Not sure. I haven’t tried it myself, but people tried to use it in Chinese, and you can’t get a full answer as it’s censored. It keeps asking you to change the topic.
DeepSeek - Yes please and thank you!
(I also second what someone said below about adding OpenRouter. That would be sick!)
Yes, please! I’d love to see DeepSeek as an available option.
I agree; yes, please, and thank you
Yes, please; I agree.
It appears appealing, however, I recommend reviewing their privacy policy. Based on my understanding, they may utilize our data for their own advantage. This could potentially be superficial. Additionally, this company is owned by Chinese interests.
We have heard you! We’re getting the wheels in motion. Interestingly, at least currently, the link to their API page is broken: https://platform.deepseek.com/
Once they get that fixed up, we’ll get started!
Thank you! Whats a general time frame after there pages gets back up? (Its fine if you dont have one)
That’s great news! But please…please…let R1 show its reasoning and don’t put it behind the three ‘•••’. That said, reasoning should be clearly separable from the final response.
I feel one reason deepseek became so popular (barring pricing) is because its reasoning tone and style leads people to anthropomorphize it more. If o1 did the same, people would like it more. Chatgpt is launching persona controls where you can specify its ‘traits’. They seem to be heading in the same direction.
Here’s 18 upvotes for Deepseek integration!
Thank you Nathaniel, thank you PickAxe!
DeepSeek is for coding and math. It is not designed for, nor does it excel at, writing.
DeepSeek R1 is now available on:
- Microsoft Azure
- GitHub
- Perplexity
- NVIDIA NIM
- Windsurf
Looking forward to using DeepSeek in our PickAxes soon
Yes! @vener I think it would be better for data retention and privacy law compliance if the API comes from U.S.-based vendors instead of directly from DeepSeek (China).
My clients/users will feel much better knowing the model is hosted in the U.S.
I’d much rather see the new o3 mini models in Pickaxe instead. There have already been serious security concerns with DeepSeek, including data leaks and strong indications of state-sponsored propaganda. In my opinion, this model simply cannot be trusted. I even installed DeepSeek on my Mac, but within an hour of testing, I got nervous and erased it. On the other hand, o3 mini has been giving me crazy good results—one of them is blazingly fast!
This should be a nobrainer