what's with the paid API requirement for open-source software?

[Njasa2k]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

[adamreading]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

When did they start charging? I’ve been using this for about a week - the api key was free to get and they didn’t ask any card details or anything to use it?

[yx9966]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

All features of ComfyUI-Copilot are free for users. The purpose of API keys is to distinguish between different users during program operation.

Additionally, for locally deployed models: Smaller models (like 7B or 14B) cannot adequately support the complex agent architecture behind ComfyUI-Copilot, which would negatively impact user experience. This approach can be referenced to some current practices like Cursor. And all the api fees which charged by third parties are paid by our group, not user of CC. Thank you for your interest in ComfyUI-Copilot! Feel free to try it out and share your feedback on its features.

[yx9966]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

When did they start charging? I’ve been using this for about a week - the api key was free to get and they didn’t ask any card details or anything to use it?

Thanks for the reply. We are currently conducting user research to enhance Copilot's features. Would it be possible to contact you for participation in this study?

[adamreading]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

When did they start charging? I’ve been using this for about a week - the api key was free to get and they didn’t ask any card details or anything to use it?

Thanks for the reply. We are currently conducting user research to enhance Copilot's features. Would it be possible to contact you for participation in this study?

100% give me a discord link or email address and I’ll get in touch.

[yx9966]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

When did they start charging? I’ve been using this for about a week - the api key was free to get and they didn’t ask any card details or anything to use it?

Thanks for the reply. We are currently conducting user research to enhance Copilot's features. Would it be possible to contact you for participation in this study?

100% give me a discord link or email address and I’ll get in touch.

shali.yx@alibaba-inc.com

[TruthSearcher]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

All features of ComfyUI-Copilot are free for users. The purpose of API keys is to distinguish between different users during program operation.

Additionally, for locally deployed models: Smaller models (like 7B or 14B) cannot adequately support the complex agent architecture behind ComfyUI-Copilot, which would negatively impact user experience. This approach can be referenced to some current practices like Cursor. And all the api fees which charged by third parties are paid by our group, not user of CC. Thank you for your interest in ComfyUI-Copilot! Feel free to try it out and share your feedback on its features.

Did you not use a 7B model in the paper?

[yx9966]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

All features of ComfyUI-Copilot are free for users. The purpose of API keys is to distinguish between different users during program operation.
Additionally, for locally deployed models: Smaller models (like 7B or 14B) cannot adequately support the complex agent architecture behind ComfyUI-Copilot, which would negatively impact user experience. This approach can be referenced to some current practices like Cursor. And all the api fees which charged by third parties are paid by our group, not user of CC. Thank you for your interest in ComfyUI-Copilot! Feel free to try it out and share your feedback on its features.

Did you not use a 7B model in the paper?

The 7B model is specific for workflow generation function, which is a tool ,not the whole agent driver.

[Njasa2k]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

All features of ComfyUI-Copilot are free for users. The purpose of API keys is to distinguish between different users during program operation.
Additionally, for locally deployed models: Smaller models (like 7B or 14B) cannot adequately support the complex agent architecture behind ComfyUI-Copilot, which would negatively impact user experience. This approach can be referenced to some current practices like Cursor. And all the api fees which charged by third parties are paid by our group, not user of CC. Thank you for your interest in ComfyUI-Copilot! Feel free to try it out and share your feedback on its features.

Did you not use a 7B model in the paper?

Yeah I feel a bit misled. I was hoping for a purely local option for the open-source community. If it was going to require internet connection, I'm not sure this will catch on.* People that use ComfyUI are not generating across the internet.

[yx9966]

@Njasa2k Got it. We have gone through multiple rounds of testing on the backend, including the accuracy of Tool Use, the accuracy of RAG, and the response time. The current open-source LLM models, except DeepSeek, are difficult to achieve the ultimate capabilities that satisfy us. Local deployment of LLM will also occupy more user GPU resources. Considering that many Comfy users run on consumer-grade graphics cards, overly large models will also occupy more user resources. The current choice is also a balanced result.
We do plan to place the LLM scheduling content in the user's Comfy-UI environment in the next version, so that users can choose the LLM model they need to interact with (it can be local, it can be free from us, or it can be directly interacting with the API of models such as openai). The implementation of this plan will take some time. Thank you very much for your suggestions.

[adamreading]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

All features of ComfyUI-Copilot are free for users. The purpose of API keys is to distinguish between different users during program operation.
Additionally, for locally deployed models: Smaller models (like 7B or 14B) cannot adequately support the complex agent architecture behind ComfyUI-Copilot, which would negatively impact user experience. This approach can be referenced to some current practices like Cursor. And all the api fees which charged by third parties are paid by our group, not user of CC. Thank you for your interest in ComfyUI-Copilot! Feel free to try it out and share your feedback on its features.

Did you not use a 7B model in the paper?

Yeah I feel a bit misled. I was hoping for a purely local option for the open-source community. If it was going to require internet connection, I'm not sure this will catch on.* People that use ComfyUI are not generating across the internet.

I don’t entirely agree with you. I’d like the option to use my local LMStudio on my 5090, but running that also takes a chunk of my imaging VRAM away. When I build workflows I generally use Groq nodes to pull in maverick or Scout, I don’t want paid API’s but I also don’t want to run locally. The speed of the current system is a bit prohibitive. The workflow generation isn’t yet available for me to test. But the idea of how it’s working is fantastic. A few tweaks and modifications and there could be a system I would have no hesitation recommending to all my Comfyui friends.

[yohaiai]

@yx9966 hello. Will you release the ComfyUI-R1 model?

[adamreading]

what's with the paid API requirement for open-source software? can't it be done with a localAI, is there any component of this software that is not open-sourced?

When did they start charging? I’ve been using this for about a week - the api key was free to get and they didn’t ask any card details or anything to use it?

Thanks for the reply. We are currently conducting user research to enhance Copilot's features. Would it be possible to contact you for participation in this study?

100% give me a discord link or email address and I’ll get in touch.

shali.yx@alibaba-inc.com

Just out of interest - You added me on discord - I came back to you with a ton of information and you never responded to the dm...

[yx9966]

@adamreading chat with you on reddit~plz check it. Sorry again for the delay of response.