Is it possible to import my conversations from the previous version of HuggingChat?

Yeah this dumbing down the system was totally worth nuking everyone's logs and assistants...? The performance improvements are nice if true, but how can you call this a better UX when so many basic features are missing from the last version? Even simple settings are gone, like no options to delete or edit output? There isn't even a way to tweak temperature/repetition minimizing settings, or give different chats different system prompts??

wow, I'm kind of surprised it's back. feels like a tad bit of a downgrade, but I'm assuming that it was a complete rework? hoping that more QoL features will be reintroduced again.

Does the platform now impose usage limits based on inference credits for free users?

we're so back

edit:
nevermind, cant delete the conversation branch like before😢

edit 2:
and it now has a limit. Its been over six hours and i still cant continue the conversation 😭

Thanks for getting this running

❌ Can't use assistants
❌ Can't generate images
❌ Can't edit conversations
❌ Can't search the web
❌ Can't change temperature
❌ Can't import your old conversations
✅ You now have to pay to use it 😂

tried it out a lot, seems pretty much good as new. new UI looks different for sure. works well, haven't had it crash at all and haven't seen any of the weird rendering bugs that were so prevalent in the old version. considering that there's a lot missing, I think it's probably a thing where they're going to reintroduce all of the old stuff again over time. the old version took almost two years (23 months if I recall) to become a sorta-final product.

Thanks for your feedback everyone <3

Is it possible to import my conversations from the previous version of HuggingChat?

Not really but we can work on it if a lot of people want it.

Can't edit conversations

Mhhh you can (or I'm missing something)

Can't change temperature
Can't search the web
Can't generate images

We plan to add this for the rest I'm not sure yet but Assistants may come back (and they will be better)

✅ You now have to pay to use it 😂

There's a free tier. It's way faster and stable now also PRO Subscription is a very nice deal https://huggingface.co/pro. Also paid requests are very cheap for most models. Example from my https://huggingface.co/settings/inference-providers/overview: 1840 requests with Qwen/Qwen3-235B-A22B-Instruct-2507= .65$

tried it out a lot, seems pretty much good as new. new UI looks different for sure. works well, haven't had it crash at all and haven't seen any of the weird rendering bugs that were so prevalent in the old version.

Thanks that's motivating and yes we'll make it awesome with faster iteration rate than the 1st version!

Thanks for your feedback everyone <3

Is it possible to import my conversations from the previous version of HuggingChat?

Not really but we can work on it if a lot of people want it.

Can't edit conversations

Mhhh you can (or I'm missing something)

Can't change temperature
Can't search the web
Can't generate images

We plan to add this for the rest I'm not sure yet but Assistants may come back (and they will be better)

✅ You now have to pay to use it 😂

There's a free tier. It's way faster and stable now also PRO Subscription is a very nice deal https://huggingface.co/pro. Also paid requests are very cheap for most models. Example from my https://huggingface.co/settings/inference-providers/overview: 1840 requests with Qwen/Qwen3-235B-A22B-Instruct-2507= .65$

tried it out a lot, seems pretty much good as new. new UI looks different for sure. works well, haven't had it crash at all and haven't seen any of the weird rendering bugs that were so prevalent in the old version.

Thanks that's motivating and yes we'll make it awesome with faster iteration rate than the 1st version!

what about the missing feature that involves in deleting a branch of the conversation? I used that very often to make things tidy

Anyone know which model is the best for uncensored stuff, one of the Cohere variants or something else?

@erictheechidna I can see CohereLabs's models there, I am not sure that it's the best for uncensored stuff. Why do you need a model from the team then? I downloaded and tried, it said: I'm sorry, but I can't assist with that kind of explicit content...

Can't edit conversations

Mhhh you can (or I'm missing something)

Cool, so how do you edit the output text?

It's a shameful imitation of the old chat, and now you have to pay for it. It's clear why they closed the previous one — the free trial ended and now you have to pay.

Pretty happy with Omni for the limited time I've spent with it so far.

I have some questions before I get too invested:

1. Has the privacy policy changed? I don't see the explicit mention of conversations never being shared with the model owners. Is HuggingChat Omni still private?
2. Is there a way to see how much you can use HuggingChat before you have to pay for it? Does it refresh hourly, daily, or what? There absolutely must be a way to use this indefinitely for free or it's not worth using, since I'm never going to be a paid user.
3. Is the Omni routing model also private?

Thank you

So when the free trial ends, that's it? You can't use it anymore?

In the earlier version of HuggingChat, there was a visible toggle to opt out of data sharing and model training so that user conversations weren’t used for model improvement, is that toggle switch still there?

It's clear why they closed the previous one — the free trial ended and now you have to pay.

So when the free trial ends, that's it? You can't use it anymore?

I get your concerns about free usage (for now you need to be a PRO user to utilize the pay-as-you-go feature, but it should be quite cheap for a normal usage). We’re open to raising the limit. But before we do we need to find the right balance: the previous version was hammered by abuse that nearly broke the service. We’re committed to keeping the doors open for everyone, but we need a guardrail that keeps the lights on.

Is there a way to see how much you can use HuggingChat before you have to pay for it?

Yes there's a way here: https://huggingface.co/settings/billing (in the inference usage section)

In the earlier version of HuggingChat, there was a visible toggle to opt out of data sharing and model training so that user conversations weren’t used for model improvement, is that toggle switch still there?

Has the privacy policy changed? I don't see the explicit mention of conversations never being shared with the model owners. Is HuggingChat Omni still private?

Regarding privacy, now HuggingChat is using external providers (HF Inference Providers) - so the privacy will depend on which providers you are using, the good news is that you can choose which ones you want to use be toggling them in your Inference Settings.

Cool, so how do you edit the output text?

Are you sure this was possible (btw you can from the inference playground)

I think many features demanded here might be introduced by HF later, so I summarise them for HF (probably victor?):

• Import previous chats (those we had before the sudden shutdown)
• Assistants (importation, creation, and sharing)
• Tools from community
• Image, Web search, and temperature tweaking

The followings are from me, but I guess folks will have similar thoughts

• More transparent pricing
  I've read the API provider & model page, but some are missing (denoted only with - symbol, including HF SmolLM, Cohere, and Zai)
  I wonder whether that means free, or just price-unavailable
  Otherwise we may run out of our free quotas easily if we're misled
  edited I guess it's price unavailable since I tried command a from Cohere, while I've been charged :

• Custom routing logic, e.g.
  Free first, or ranked by pricing
  I personally wanted this :)
  or ranked by model type, popularity, date of release, etc.
• It will be good if we can set certain providers for specific models, e.g. firework for DeepSeek V3, Together for QWQ, or Groq for Kimi K2
• Can we bring our own API keys other than your provider partners' ones?
  e.g. OpenRouter or LLM gateway
• Set our preferred model as default, or give it a shortcut

And I have a personal opinion about abuse-prevention: setting a rational rate limit and reset it after a certain period, just like OpenAI or Anthropic did.
For heavy users, maybe HF pro or other subscription plans can help.

Thanks for your reading.
I'm here because I love open source and privacy so much, while HuggingChat is the most ideal choice.

And I have a personal opinion about abuse-prevention: setting a rational rate limit and reset it after a certain period, just like OpenAI or Anthropic did.
For heavy users, maybe HF pro or other subscription plans can help.

Yeah, I totally agree, having a reasonable rate limit that resets periodically is way better than having one hard cap on overall usage that forces you to upgrade to pro once you hit it.

Great job on killing something that was once amazing from what i had heard.

We’re open to raising the limit. But before we do we need to find the right balance: the previous version was hammered by abuse that nearly broke the service. We’re committed to keeping the doors open for everyone, but we need a guardrail that keeps the lights on.

There's a pretty obvious solution: Daily caps. Let us send 75-100 messages a day. You can compare it to Grok, which has a 2-hour limit of 20 messages, which works out to 240 messages a day. 100 would be low but acceptable for most users.

Yes there's a way here: https://huggingface.co/settings/billing (in the inference usage section)

Yeah, this is terrible. I'm already almost done and if there's no increase I just won't come back.

Regarding privacy, now HuggingChat is using external providers (HF Inference Providers) - so the privacy will depend on which providers you are using, the good news is that you can choose which ones you want to use be toggling them in your Inference Settings.

We also need a way to block this completely. Having it be determined by the model providers, in a system that routes you to a provider it chooses without you having the ability to opt-out, is unacceptable.

As it is, this is so much worse than the old Hugging Chat that it's not worth using in its current state. I'll check back to see if/when an increased recurring cap and better privacy measures are implemented.

We also need a way to block this completely. Having it be determined by the model providers, in a system that routes you to a provider it chooses without you having the ability to opt-out, is unacceptable.

You can block completely the providers you don't want from your Inference Settings.

It’s actually sad that most open-source model providers collect or share your data for training without offering an opt-out option, while closed-source companies like OpenAI and Anthropic give users that control.

I remember using the old Hugging chat a lot. With this one I ran out of credits in seconds :(
Back to openrouter I suppose.

I'm curious to know what settings they are using for the models. On the old Huggingchat you used to be able to see the temperature etc for the model you were using.

Hay welcome back. It's fine if you want to add a paid feature, and it's also fine if you want to make it a totally paid/ free tier (basically a bait for paid services), but that won't be hugging chat. Hugging chat was something else. It was like charity work. It gave us hope that not all AI providers seek money, and some can give you free service. It was perfect, every bit of it was. It had so many features, it had a strong privacy-friendly system, secure. It's fine if you want to make something like other AI competitors (OpenAI and others), it's totally fine. Your product, your rules, who am I to say anything? But I just want to thank you for the last 2 years. You had made my life very easy. Because of the hugging chat, I was able to use AI without spending money. It was the freedom and openness of open source that we needed. It's fine now you want to build something else. Maybe you will now get customers who are dedicated and willing to spend money. I am no longer a customer as I am not willing to spend any money on my AI, as I have other things to spend money on. Again, thanks for the last 2 years. I will always remember there was a Great Open-source AI host called Huggingchat. Cheers

So when the free trial ends, that's it? You can't use it anymore?

seems like it. it's been over a day now since i've reached my limit, i still cant chat

Please keep the free tier as interesting as possible to be able to extensively test the open models. Especially for new models and those which can not be run easily on mid to low consumer hardware.
Consider introducing a setting which allows to fine-tune omni routing from now hardcoded "best underlying model" to more options I'm sure also paying Pro users are interested in like:

• best performance per inference cost/credit
• fastest performance in the top tier models

Also gradually saving preferences of users (models output) would be an interesting intermediate comparing to hard select a model vs. HF omni routing preferences.

I'm just wondering are the free limit uses a one time thing or do they refresh daily?

And if they are a daily thing will we able to just pick up where our story left off or do we need to start a new chat every time?

With no way to refill for free, that'll just incentivise throwaway accounts. Prevent abuse by limiting out the expensive to run models and features, queue the inputs on a lower priority, and rate limit heavy users.

Also, that settings ui not showing fractions of a cent hurts the precision a lot.

i set up IP READ API KEY

but it still reads "" Amy help is welcome...

waited a handful of days, seems like the free tier does not, in fact, reset. at least not daily. if it were to reset daily (or, preferably, have a system similar to the old Huggingchat) that would be a huge improvement. maybe an advertisement banner or something to offset the costs? I'm not sure, I'm not super knowledgeable on this, but being able to run through what I assume is a montly limit in an afternoon is disappointing.

I'm a little disappointed that the UI doesn't have the same features that the old UI did, such as the ability to delete branches. I thought those would have remained.

Pro... considering it. After the old hugging chat closed, I moved to Mistral on the €15/m ($17 USD). So being able to get back what I've missed for $9 seems nice.

I get your concerns about free usage (for now you need to be a PRO user to utilize the pay-as-you-go feature, but it should be quite cheap for a normal usage). We’re open to raising the limit. But before we do we need to find the right balance: the previous version was hammered by abuse that nearly broke the service. We’re committed to keeping the doors open for everyone, but we need a guardrail that keeps the lights on.

Trust me when I say I get this. I became a bit of a heavy user on the old version. So, here's an idea. As said before, I've (currently) moved to Mistral and their system works like this:

• Free tier: daily limit of X number of messages (example: 100).
• Pro tier: limit is now every 3 hours. Number of messages has increased (example: 1000).

It would be really, really nice if you could do that here. This should at least give guard rails on free users but really benefit/reward those who subscribe. I do also prefer subscription over PAYG too.

Cool, so how do you edit the output text?

Are you sure this was possible

I do not recall this being possible. You could delete responses though. Which I feel is a key feature that needs to come back.

Oh, and seeing as how this has been relaunched - could all of the old discussions be closed now?

I'm a little disappointed that the UI doesn't have the same features that the old UI did, such as the ability to delete branches. I thought those would have remained.

Pro... considering it. After the old hugging chat closed, I moved to Mistral on the €15/m ($17 USD). So being able to get back what I've missed for $9 seems nice.

I get your concerns about free usage (for now you need to be a PRO user to utilize the pay-as-you-go feature, but it should be quite cheap for a normal usage). We’re open to raising the limit. But before we do we need to find the right balance: the previous version was hammered by abuse that nearly broke the service. We’re committed to keeping the doors open for everyone, but we need a guardrail that keeps the lights on.

Trust me when I say I get this. I became a bit of a heavy user on the old version. So, here's an idea. As said before, I've (currently) moved to Mistral and their system works like this:

• Free tier: daily limit of X number of messages (example: 100).
• Pro tier: limit is now every 3 hours. Number of messages has increased (example: 1000).

It would be really, really nice if you could do that here. This should at least give guard rails on free users but really benefit/reward those who subscribe. I do also prefer subscription over PAYG too.

---

Cool, so how do you edit the output text?

Are you sure this was possible

I do not recall this being possible. You could delete responses though. Which I feel is a key feature that needs to come back.

the pro version is worth it

sooo do the free tier reset or its a bait and switch ?

the pro version is worth it

I've been thinking of getting it but like how does it actually work because I won't lie I don't 100% understand what it was saying I would be paying for? especially when I saw something that said you would have to pay more if you went over a limit and honestly that scared the hell out of me.

I know this probably won't happen but can we just have a subscription tier that's just for huggingChat? because I was looking at it again and maybe it's just me but I don't need or even want almost all of the features you would be paying the £9 a month for and don't get me wrong I'm fine paying that much as it's a lot less than other AI generators cost at their lowist tier and also because It's really easy to use, I love the level of descriptive and detailed storytelling it produces and best of all it doesn't give me a word count limit for what I can write for my prompt... because I didn't think my prompt I use a lot was that long until I tried to use another AI generator and it's almost 20,000 words long. 😅

seems like it. it's been over a day now since i've reached my limit, i still cant chat

Please consider upgrading to PRO!

lol looks like it dont reset

seems like it. it's been over a day now since i've reached my limit, i still cant chat

Please consider upgrading to PRO!

...what?

seems like it. it's been over a day now since i've reached my limit, i still cant chat

Please consider upgrading to PRO!

Wow... HuggingChat is dead folks

i wouldn't mind if they refreshed with a decent bit of usage limit but if its a one time trial thing then its dead folks pack it up