Private AI

Nano GPT logo
NanoGPT
Nano GPT logo
NanoGPT
Nano GPT logo
NanoGPT
Search
 
 
 
 
All conversations

Create

Chat
Studio

Explore

Models
Pricing
AI Apps
Blog

Library

Gallery
Workspace
Usage & billing
API
Suggestions and bugs
Use cases

|

Terms

|

Privacy

|

Refunds
Usage & billing
Teams
Referrals
Presets
Settings
FAQPress & mediaBrand assetsBenchmarksNano RPCEarn crypto
164 community ideas completed

Suggestions and bugs

Share ideas and upvote the ones you want most. Bug reports stay private between you and our support team.

ideas shared
296
open ideas
105
votes cast
1,308
Community board

What people are asking for

please add a higher tier of sub

OpenUpdated Apr 28, 2026

if the 2x token use on model name is not possible, please add a double price tier of sub where 2 x token models are 1x. can be the same limits as the current, but it would eliminate the uncertainty of selecting models, I am sure many of us will happily pay the extra.

More Abliterated/Uncensored Models

CompletedUpdated Nov 11, 2026

It would be perfect if you can add more strong uncensored models for coding. GLM 5.3 flash uncensored is good but it is the only one that i can use for coding. I would like to see deepseek flash or other strong uncensored models if it is possible.

Add kimi 3.1 Pro Uncensored

OpenUpdated Nov 7, 2026

Pls add kimi 3.1 pro uncensored, it's best ai for cybesec as i know

Remove CloudFlare

OpenUpdated Feb 15, 20261 reply

I noticed your site and API are behind the Cloudflare proxy/CDN. Cloudflare inspects and routes all visitor traffic, which can expose users’ IP addresses and request data to a third party and centralizes metadata that could be retained or accessed. For better privacy and to avoid third‑party interception of user requests, please consider removing the Cloudflare proxy (or configure DNS only / self‑hosted CDN) so visitors connect directly to your origin servers. Thank you.

NanoGPT teamFeb 15, 2026

You’re right to flag the privacy implications of reverse proxies/CDNs and we’re evaluating options to reduce third-party exposure where feasible.

For clarity: NanoGPT is not currently behind a Cloudflare reverse proxy. We use our hosting provider’s global edge network (running on AWS) for performance and DDoS protection, which means AWS/our edge provider may still process similar connection metadata.

Add support for Refuseless (refuseless.com) models and provider

OpenUpdated Nov 7, 2026

Please consider integrating Refuseless (https://refuseless.com/) into NanoGPT. Adding their models would be a great fit for users who need uncensored, unfiltered responses with minimal false-positive refusals—especially for creative writing, roleplay, security research, and complex technical tasks. Since Refuseless provides standard OpenAI-compatible API endpoints, integrating them as a provider on NanoGPT should be straightforward and would expand the catalog of unrestricted models available on the platform

Show release date in model overview

OpenUpdated May 21, 2026

It would be nice to see at a glance when a model was released.

Custom HuggingFace support

AcceptedUpdated Jan 29, 20262 replies

Your application already supports over 500 LLMs, and the Custom CivitAI update got me thinking - it would be incredible if users could access HuggingFace models through NanoGPT. Because RAM and GPU prices are rising, an AI rig is impossible for the average consumer, myself included. Adding this feature would be perfect because users could access more models and use them without getting expensive hardware. The basic idea would be to find a HF URL, like https://huggingface.co/mradermacher/Huihui-GLM-4.7-Flash-abliterated-i1-GGUF (a currently inaccessible model) that contains GGUFs and then to paste and submit it. Like Custom CivitAI, the LLM would have to be resolved to be used, but once finished you could start using it. Likely all users would enjoy this feature.

NanoGPT teamJan 29, 2026

Thanks! We'd love to add something like this but as far as we know this is not possible to add in a cheap way - as in while this is possible for image models there is no "load on demand" for text models. Maybe we are wrong - if you did find it somewhere we'd definitely love to know! - Milan

Continue generating text

CompletedUpdated Nov 11, 2026

Sometimes a reasoning model gets interrupted, and does not finish the task. Currently i make them continue by sending a message that just says "continue". Maybe it's better to provide a separate button to continue generating.

Provider selection for proprietary models

OpenUpdated Oct 27, 2026

Even for Claude and Gemini models, there are a few different providers available and OpenRouter allows to choose between them. This is useful, because for example Google Vertex tends to be less censored than AI Studio or Anthropic (or at the very least it allows users to judge if a specific provider works better for their needs). They also have different data retention policies which one could compare if it was transparent who the requests are routed to. I don't know why the option to set provider preferences is exclusive to open source models. Additionally, the reasoning for not showing the provider used for each request on auto routing does not seem to apply here, because the prices billed for these models are apparently identical with all providers (on the consumer's end, which is what matters regarding the possible disclosure of discounts). So I'd like to be able to specify the provider for Gemini and Claude models and for auto routing to report which provider served these models for each request.

Model -> providor settings url

OpenUpdated Nov 10, 2026

I want an url to receive info from every model available in nanoGPT.com, so I can receive all model settings info at once.. that a provider can make. where eg: reasoning buttons are displayed, and eg: { "temperature": {"min": 0, "max": 2}, "top_p": {"min": 0, "max": 1}, "top_k": {"supported": false}, "min_p": {"supported": true}, "repetition_penalty": {"min": 1, "max": 2}, "reasoning_efforts": ["low", "medium", "high"] } I need that

Auto routing usage and extra information

OpenUpdated Oct 29, 2026

When auto is used the usage logs do not show which provider the request was routed to. It would be nice to be shown which provider was picked. Also, when auto routing is the only option available, the provider should be disclosed.

Pro / coding Subscription

OpenUpdated Feb 26, 20261 reply

As open source models are becoming an alternative for commercial models, especially for coding, the reliability needs are also increasing. The subscription provides a good way to have access to such models. For normal chat completion requests, "cheap" providers (your backend providers) are usually sufficient. But for coding, high reliability is needed (reasoning, tool calling). Some providers apply settings to the models that make them nearly impossible to use for coding. Therefore I suggest a pro or coding subscription with the same / similar limits like the normal subscription, but with providers who don't heavily modify the model settings (usually the model dev providers).

NanoGPT teamFeb 26, 2026

We are unlikely to do this or if we do it would be much more expensive - frankly we'd suggest looking at other providers for this because it's not something we are likely to do in the short term.

Include open source TEE models in subscription

OpenUpdated Jan 28, 20261 reply

They're open source so it would make sense.

NanoGPT teamJan 28, 2026

We want to do this and hopefully can do so longer term, but for now there just are not enough providers yet that do TEE to make this economically viable. TEE models also tend to be more expensive in general, so we'd likely have to charge more for this.

Real-Time Model & Provider Status Dashboard

OpenUpdated Oct 30, 2026

Hi NanoGPT team! I’d love to see a real-time model and provider status dashboard, similar to https://rs.igx.kr. As a subscription user who frequently switches between models, it’s difficult to know which models are currently working reliably without testing them myself. It would be incredibly helpful to see: * Model and provider availability * Recent success and error rates * Time to first token (TTFT) * Generation speed (tokens/sec) * Recent performance trends (5–15 minutes) Even a simple Healthy / Degraded / Unavailable indicator would be useful. This would help users choose the best-performing model before starting a conversation, rather than discovering problems after sending requests. Thank you for considering this!

Add Suno AI

OpenUpdated Dec 8, 2025

It would be great to have suno available for music generation in nano-gpt. I love your work!

Increased weekly limits for long-term subscribers

Not plannedUpdated Sep 27, 20261 reply

I read situation about those 3-5% users who caused abuse of service and unlimited tokens, and understand introduction of weekly limits fo 30mil tokens. But even among those 95% people can be few % of loyal users who may need extra tokens per week (especially when using agents or reasoning models). I would like to suggest a feature to introduce weekly limit size depending on how long user has been subscribed. It may be flat bonus after X months (i.e. 60m after 3-6 months) or incremental (i.e. +3-5mil with each month). Kinda a reward system for loyalty. And abusers usually will create a lot of new accounts and probably would not patient enough to cause issues.

NanoGPT teamSep 6, 2026

Hi,

Sorry, but we do not intend to do this. It essentially just makes the subscription even more unprofitable for us, sorry.

Feature: compress chat

OpenUpdated Mar 19, 2026

It would be cool to have some kind of button in web chat to compress/summarize whole chat. Context Memory exists, but sometimes you just need this one very unexpectedly long chat to be compressed to proceed with it after it hit context limit

Disclose providers for open-source models

CompletedUpdated Sep 27, 20261 reply

I understand that you rely on many different providers for models, including open-source models in the subscription. However, it isn't clearly written anywhere which models are hosted by which providers. (Or maybe I'm just not looking hard enough.) Any chance you could disclose this information somewhere on the site? Like, "Primary provider for X is Fireworks, fallback Chutes", as an example. Adding a note about it for the "Available Models" list under the "</> API" tab would feel intuitive, for example.

NanoGPT teamJan 11, 2026

Hi - we have added provider selection for (most of) the open source models now, so you can select your provider of choice.

It's also still possible to use the "auto" routing still, which remains the same as it was before. For that, we do not display the provider and likely will not.

API Key Usage Control

CompletedUpdated Feb 25, 20261 reply

Allow finer control of API key usage, including daily RPD and daily NANO consumption allowed.

NanoGPT teamFeb 25, 2026

It is now also possible to set a daily token limit

Add Paypal as payment option

AcceptedUpdated Dec 11, 20251 reply

I'm primarily using PayPal, so having it as a payment option would be nice

NanoGPT teamDec 11, 2025

We're working on this but this is proving quite a bit more difficult than we had hoped. Sorry - we'd love to add this in ourselves as well! - Milan

Clarify the privacy/security of using NanoGPT storage/sync

OpenUpdated Feb 3, 2026

An eventual audit might be helpful too. And publishing the source code.

Simplified API Pricing Description

OpenUpdated Nov 25, 2025

I want to make the API pricing easier to understand. This includes showing the price for image inputs and for video generation with or without audio. Some models do not have pricing, so I also want to clearly show which features each model can access, such as whether image input is allowed.

MiMo V2.5 Pro open-sourced and not yet is subscription

CompletedUpdated Nov 11, 20262 replies

title says it all, wondering if we'll get it in the sub now that it's opensourced, since we have glm 5.1 but not yet mimo, wanted to test that it's allegedly the current top dog

NanoGPT teamMay 3, 2026

Hi,

So we'd love to, but there are no open source providers actually hosting the model so far. So it is open source, but no one seems very enthusiastic about offering it to us.

Kind regards,

Milan

Move model selector location away from input field on Mobile

OpenUpdated Nov 13, 2026

UI Request: Recommend moving the “model selection” dropdown from the bottom to the TOP of the window to avoid accidental selection. Problem: The location of the “model selection” drop down at the bottom of the screen on iOS/iPadOS 27 makes it prone for “accidental selection” when a user is attempting to select the ”Ask anything” “text entry” field. (Observed on IPad and iPhone) Love the site. Use it extensively. Thank you!

Google safety filters passthrough

CompletedUpdated Nov 11, 20262 replies

Google vertex and ai studio accept lowering safety filters. You can technically set them off like so on Openrouter: ``` "provider": { "order": [ "Google" ], "allow_fallbacks": false }, "safety_settings": [ { "category": "HARM_CATEGORY_HARASSMENT", "threshold": "OFF" }, { "category": "HARM_CATEGORY_HATE_SPEECH", "threshold": "OFF" }, { "category": "HARM_CATEGORY_SEXUALLY_EXPLICIT", "threshold": "OFF" }, { "category": "HARM_CATEGORY_DANGEROUS_CONTENT", "threshold": "OFF" } ], ``` But they really are still engaged in some capacity, but it'd be nice to have just for posterity. I think the 4 above are for ai studio specifically, and more are available for vertex. https://docs.cloud.google.com/gemini-enterprise-agent-platform/reference/rest/v1/HarmCategory?hl=en Thanks.

MilanNov 11, 2026

Thanks for the suggestion! Gemini on NanoGPT already runs with all adjustable safety filters off. Where the provider accepts the setting we send OFF for every category, and on the newer Gemini models (2.5 and later, including 3.x and 4) Off is also Google's default. We tested it: sending "OFF" gives the same result as sending nothing, and the only thing passing these settings through would allow is making filtering stricter. The refusals you still see come from the model itself and from Google's fixed filters, which no setting can change.

Clarify provider data retention agreements

CompletedUpdated Sep 27, 20261 reply

I remember that NanoGPT used to advertise "zero data retention" agreements with their LLM providers. At some point this language has been removed (or I'm misremembering) and replaced with "minimum data retention", which is a far more nebulous term, though even "zero data retention" does not have a standardized definition among providers (especially regarding abuse detection). Can you disclose which providers you have special data retention agreements with and what those terms are? It would go a long way to help the transparency of this platform.

NanoGPT teamSep 6, 2026

We’ve added provider-level privacy information to the model provider lists, including whether providers retain prompts or may use them for training, along with links to their policies. We also offer Zero Data Retention routing for text requests.

ZDR reflects provider commitments; it isn’t an independent audit of their systems. Our privacy explanation is at https://nano-gpt.com/privacy.

Add Ornith 1.5 397B variant

OpenUpdated Oct 27, 2026

I noticed that the 35B and 122B variants are already here, so it would be good to have the biggest model too

Anonymous vouchers

OpenUpdated Sep 9, 20261 reply

Hello NanoGPT team! A suggestion I have is to provide vouchers like Mullvad does so people can deposit money anonymously without needing crypto. I didn't want to deal with the tax headache of cryptocurrencies, so I used the stripe process to add funds. I was under the impression that personal information would not be knowable by NanoGPT, but I can see my personal information under billing so you surely must have access, right? Vouchers would enable a clean separation for those like myself which do not want to rely on trust.

NanoGPT teamSep 9, 2026

We’ve reached out to services such as Bitrefill about offering NanoGPT vouchers, but unfortunately haven’t heard back yet. We don’t have an availability date to share at this point.

We really do want to minimize the information that we see, but unfortunately, Stripe does not actually have a setting or option to see even less than we do now. We do see, indeed, the name you fill in and such there (within Stripe, it does not reach our own database).

Bridge discord to XMPP or Matrix

OpenUpdated Mar 23, 2026

The discord room should be bridged to an XMPP or Matrix group chat. Lots of privacy and freedom conscious people stay away from user-hostile software & company like discord, and thus, they are unable to participate in the user community discussion with the nano-gpt developers.

More filter options

CompletedUpdated Nov 15, 20262 replies

More filter options for models (all of them) would be useful, such as "can be coaxed easily into NSFW", "does not guarantee privacy", etc...

HuggiNov 15, 2026

We’ve added more filters to the text models page (nano-gpt.com/models/text), under More filters: "Open source only", a price ceiling ($1, $5 or $15 or less per 1M tokens), and a Creators filter to show only, or hide, models from chosen companies. They also work with search and are saved in the page link. We still don’t have a filter for how easily a model goes NSFW.

Show TokensPerSecond for each model?

CompletedUpdated Feb 6, 20261 reply

Is there a way to see supported TPS for each model like OpenRouter displays?

NanoGPT teamFeb 6, 2026

We now do this for most (if not all?) models. - Milan

Make a .onion address

CompletedUpdated Aug 24, 20261 reply

Would be nice if there was an anonymous way to access this via Tor. Probably would have to disable the free model for that, though.

NanoGPT teamAug 24, 2026

🧅 NanoGPT is now available as an onion service

We are still testing the implementation, but you can now access the NanoGPT API directly through the Tor network.

Onion address

http://nanogpt5wwal7shnauuilouj3gaewmk7lskqxdmxlbfz5ho4tdqeh2id.onion

API base URL

http://nanogpt5wwal7shnauuilouj3gaewmk7lskqxdmxlbfz5ho4tdqeh2id.onion/api/v1

Your existing NanoGPT API key works as usual. Connections to the onion service remain within the Tor network and do not pass through Tor exit nodes.

Please test it with your usual clients, models, and streaming requests. If something breaks, let us know which client and endpoint you used.

Add more API settings

CompletedUpdated Nov 21, 2025

Please consider adding generation settings such as Min-P, this is really helpful since most of prompts and presets for roleplay comes along with generation settings and its bad that nanoGPT API does not support Min-P setting.

NanoGPT team

Hi! We already supported most of those but didn't have it in the documentation properly. https://docs.nano-gpt.com/api-reference/endpoint/chat-completion has now been updated!

Opencode

CompletedUpdated Nov 11, 20261 reply

Can you contact the OpenCode team so we can use all our subscription models, as they aren't currently listed.

NanoGPT teamFeb 6, 2026

Could it be that you are using api.nano-gpt.com? As far as we know they just use the v1/models endpoint which should return all. But the URL should be nano-gpt.com/api/v1. - Milan

Attachment 1

Prompt Caching for more models

CompletedUpdated Sep 27, 20262 replies

Many providers do offer prompt caching nowadays and utilizing that yields great discounts. Unfortunately, through NanoGPT, we can't benefit from those (except with Claude, which is the most inconvenient to implement). That remains a big advantage of OpenRouter. For me personally, the most interesting providers are Moonshot AI, Google (Gemini) and DeepSeek. I am aware there are some suggestions like this already, but they can't be voted on, so this is my way of expressing demand. I'm also aware this might hurt your profits, if you only pay discounted rates while you charge base costs, but that would also mean the pricing isn't fair.

NanoGPT teamFeb 26, 2026

We now haev this for Gemini as well - Milan

Add Qwen3 Embedding 0.6B to Subscription Models

Not plannedUpdated Jan 15, 20261 reply

Qwen3 Embedding 0.6B A powerful and cheap Embedding model that is also open source and available here: https://huggingface.co/Qwen/Qwen3-Embedding-0.6B, it would also be nice to obtain vectors without spending API requests.

NanoGPT teamJan 15, 2026

Sorry, but we're not going to add embedding models to the subscription. - Milan

Model Date

CompletedUpdated Dec 8, 2025

I think it would be a good addition to include details about when a model was last updated or uploaded. Why? For example, the recent appearance of the Mistral Large 3 model. It appears in the available models list, but I'm not sure if it's the new version or an older one. While it might have "2512" at the end of the model name, this doesn't guarantee a specific date, as the other existing models don't have anything like that.

NanoGPT team

We've now also added the date-added into the hover-over description, as well as in all other places that could be relevant!

Mobile reading mode and compact layout to show more response text

CompletedUpdated Nov 11, 2026

Please add more options to make NanoGPT’s mobile interface cleaner and show more of each response at once, especially when reading long answers. On my phone, the prompt box, top controls, message information, action buttons and spacing take up a substantial amount of the screen. I’d love an optional layout that gives more of that space to the actual conversation. Some suggestions, in order of priority: 1. Reading / focus mode A toggle that hides the prompt box, top toolbar and secondary controls while reading. Keep a small, easy-to-reach button to restore the controls or reply, without losing the current reading position. 2. Compact layout for everyday chatting Reduce unnecessary padding, large gaps between messages and oversized controls. Let the prompt box collapse into a slim reply bar when it isn’t being used, expanding when tapped. 3. Less clutter around responses Put secondary actions and detailed information, such as timestamps and cost breakdowns behind a menu or expandable section. Allow suggested follow-up buttons and nonessential banners to be hidden while reading. Copying, selecting text and opening sources should remain easy. 4. Adjustable text and spacing Offer font size, a few readable font choices, line spacing and paragraph spacing settings. Keep layout density separate from font size so users can reduce wasted space while keeping comfortably readable text. 5. Better use of screen width Offer narrower side margins and reduce excessive indentation in lists on small screens. Wide tables and code blocks should scroll within their own area without making the whole conversation scroll sideways. Please remember these preferences across chats and sessions. The goal is to display more useful text with less scrolling while keeping the app comfortable to read and easy to operate. If this needs to be introduced gradually, a reading-mode toggle and collapsible prompt box would be the most useful starting point.

zdr/EU/provider preferences need to be enforced in model list requests, not returning or mentioning unusable models

CompletedUpdated Nov 11, 2026

my tooling chokes regularly due to the fact that it's attempting to use models that are disallowed by my account and its keys. the model lists returned should be correctly and reliably pruned to avoid this class of errors that causes me to create custom code specifically for nano-gpt to be usable. i ran into this issue in multiple harnesses.

Separate category for decision models

CompletedUpdated Nov 11, 20261 reply

I use NanoGPT with my OpenClaw-style personal agent, and the model list is clogged with decision models in the text category, making it difficult to browse. I think decision models deserve their own category, since while they do emit text, the structure of that text makes it a different enough task to warrant it.

MilanNov 2, 2026

Hi, thanks for the suggestion. We’ve separated decision models from the default text-model lists, including the personalized API list, so they no longer clutter the list used by chat clients and agents.

They’re available through /api/v1/decision-models. You can also add output_modalities=decisions to the existing model-list endpoints to list only decision models, or output_modalities=text,decisions to include both.

Kind regards,
Milan

Allow us to see and select providers for image generation models

OpenUpdated Oct 27, 2026

I was just thinking that it was weird that we can see and control providers for text models, but not for image generation ones.

More control over visible list of models, decision models

CompletedUpdated Nov 11, 2026

The list of decision models is growing (and others, almost 30 total now). It would be nice to be able to toggle them off/on, either individually, or even just as a whole section (and have it apply to API of course). Why? So the model list remains clean and tidy. For me, the text model "visible list" filtering is important (I only enable around 30 for now), otherwise there are hundreds to manage/disable (via other software like Open WebUI). Thank you for all the work on NanoGPT (and this suggestion area!), it's been a pleasure.

one of the tools being able to turn code into zip files

CompletedUpdated Nov 11, 2026

i want it be able to compile code together rather than just manually separating the code and then relinking it to the correct pages while also having to compile them up myself

In Browser Python MCP using pyodide

OpenUpdated Oct 27, 2026

Would be super useful for gauging a model's python skills, as well as providing it an environment to work iteratively on some python (compared to in chat web apps which are single shot files that provide no feedback to the model). An old project for this exists on GitHub at yonaka15/mcp-pyodide (compatible with the browser with some work). Otherwise, the base pyodide project is actively developed and mature.

Play Store App Store?

OpenUpdated Oct 27, 2026

Would be nice to have a playstore version of Nano on mobile

Add nanoGPT MCP to chat function

OpenUpdated Oct 27, 2026

Integrate the NanoGPT MCP with the chat function to enable the use of its tools on the NanoGPT website.

abliterated tinfoil models

OpenUpdated Oct 27, 2026

is it possible to get some abliterated / obliterated or uncensored tinfoil models?

Hi, I have two questions about Wan 3.0 Reference-to-Video on NanoGPT:

CompletedUpdated Oct 15, 20261 reply

I've noticed that the cost has increased. Previously, reference video inputs didn't appear to incur any charges, but now they do. Has the pricing policy changed? If so, when did this change take effect, and how are reference video inputs now billed? The output quality also seems to have declined compared with earlier results. Have there been any recent changes to the model version, provider, or generation settings that could explain this? Thanks for clarifying.

NanoGPT teamOct 15, 2026

Thanks for flagging this. On Oct 14 we started billing reference-video seconds at the output rate, because our provider charges us for them. We've now also lowered Wan 3.0's per-second price to $0.05 (480p) / $0.10 (720p) / $0.20 (1080p), so your 15 s clips with 15 s of reference video cost about $1.43 instead of $2.00. The Generate button shows how many output and reference seconds you're billed for. We haven't changed the model, provider or generation settings, so the quality difference isn't from our side.

Add TEE for some uncensored models

OpenUpdated Apr 9, 20261 reply

For those who'd like more privacy with one of those models

NanoGPT teamApr 9, 2026

We'd love to do this but there are few providers that host TEE models, and they genrally want to host only the most used models. The uncensored ones are not used that much, so most don't host them unfortunately. - Milan

More Usage Information

CompletedUpdated Sep 27, 20262 replies

As a user, I'd like to see more detailed information for each log in the Usage page, including provider, latency, throughput of each log/usage.

NanoGPT teamSep 9, 2026

We’ve added more detail to individual Usage entries. Open an entry to see time to first token and full-response latency when those measurements are available, alongside its billing details. We also display the provider, except if you use auto routing.

Don't see your idea?

Suggest it and let the community vote on it.