Asking AI to fix one part of an answer usually gets you a whole new answer.
Now you can select the part that's off, say what should change, and only that part is rewritten. You see exactly what changed, and the rest stays as it was.
Works on emails, code, plans, recipes, anything.
Improve is live in NanoGPT chat for everyone.
5
198
One image to a textured 3D mesh.
Idol Max and Max Turbo are on NanoGPT: 30,000-triangle GLB, UVs and 2K textures, with optional PBR and character rigging.
This is a real Idol Max output.
nano-gpt.com/studio/3d?model…
1
277
Text to a low-poly 3D asset.
Hyper3D Low Poly also works from 1–5 images or simplifies an existing mesh. Choose 200–10,000 target faces and export GLB, FBX, OBJ, STL or USDZ.
This is a real text-to-3D output.
nano-gpt.com/studio/3d?model…
1
4
329
GPT-6.2 Astra and Sol are on NanoGPT.
Text + image input, tool calling, structured output, and reasoning from low to max.
This little game was made with one Sol API call.
nano-gpt.com/chat?model=open…
1
6
326
AI answers are long. Your questions usually aren't.
Highlight the bit you care about → Cite it or spin off a side question. No more "regarding point 3, the second sentence…"
Live now in NanoGPT chat 👇
Clover 2 by Kitani is now available on NanoGPT and is free for the first week.
Clover 2 is a MoLE (Mixture of LoRA Experts) model focused on roleplay - less repetitive than most alternatives. Give it a try!
Free until November 24, 2026 03:33 UTC.
7
474
Seedance 2.5 Flash makes 4 to 30 second videos with synchronized sound, from text, start and end frames, or visual references. It can also edit an existing clip with a prompt.
This 5-second 720p clip came from one text prompt and cost $1.20. Sound on.
nano-gpt.com/studio/video?mo…
9
645
Iris-Z is live in NanoGPT Studio: one request gives you an image and a matching depth map.
It can also estimate depth from your own photo, or turn a depth map into a new image.
$0.14 per image with depth, $0.10 without. Iris-Z Small starts at $0.03.
nano-gpt.com/studio/image?mo…
4
290
Nano-GPT retweeted
Replying to @NanoGPTcom
@NanoGPTcom gift cards are now available in the TBC app!
NanoGPT gives you private access to AI, with every model available.
Chat with ChatGPT, Claude, Gemini, DeepSeek and more. Create images and video, or build with the API. Everything is paid from one prepaid balance, with no subscription required.
NanoGPT gift cards include 3% bonus credit. That means a $100 gift card adds $103 to your NanoGPT balance!
Ousia 1 by Hyperresearch is on NanoGPT as a research beta. Ask a hard question: it searches the web, reads the pages and answers with numbered citations you can check.
$2 in / $10 out per 1M tokens, about 2 to 12 cents a question in our tests.
nano-gpt.com/conversation?mo…
3
280
Inkling 2 Small TEE is live: Thinking Machines Lab's open-weight model, running in a TEE via Tinfoil.
594B total, 17B active, 524K context. It scores 31 on Artificial Analysis' Intelligence Index.
$0.30 in / $1.00 out per 1M tokens. Demo of what it built below.
3
452
Qwen 4 Max is now on NanoGPT.
Alibaba's flagship for coding, data analysis and long agent runs. Text, image and video input, a 1M-token context, reasoning from none to extra-high, tool calling and structured output.
nano-gpt.com/conversation?mo…
📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery!
Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window.
Highlights:
- Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation.
- Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables.
- Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery.
- $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%.
The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder.
Start building today! 🚀
- Blog: qwen.ai/blog?id=qwen4-max
- API: qwencloud.com/models/qwen4-m…
- Token Plan: qwencloud.com/pricing/token-…
More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀
1
13
484
Hyper3D Rigging is live on NanoGPT.
Upload a 3D character and get it back with a skeleton and skin weights, geometry and textures untouched. Add stock or custom animation clips, then export GLB, FBX or USDZ.
$0.20 per request.
nano-gpt.com/studio/3d?model…
Replying to @DeemosTech
Try HYPER3D Rigging: hyper3d.ai/
Bring your own 3D models and import custom animations.
Add or remove bones with natural language, then fine-tune the rig and motion with interactive controls.
#Rigging
2
314
Pay for AI from your TON wallet 💎
Top up with USDT from any TON wallet, or pay with Gram straight from Telegram in one tap.
Every AI model. Private. Pay as you go.
4
252
Nano-GPT retweeted
My local qwen 3.5 35b model is slow, but capable
So I gave it access to @NanoGPTcom via API key, so it can tap into more powerful models that it can coordinate
Now it’s flying 😮
Don't want to connect Gmail or Outlook to AI?
Now you don't have to. NanoGPT gives you your own email address: forward any email to it, then ask about it in any chat.
Encrypted when it arrives. Only your passkey can open it.
Set it up in Workspace → Incoming email, or in a chat choose + → Email
Shot it vertically. Need it wide?
LTX-2.5 Reframe is now on NanoGPT. It generates beyond the edges of your video, keeping the whole original shot visible. Five aspect ratios, 720p or 1080p.
Clips up to 60 seconds.
nano-gpt.com/studio?model=li…
5
627
The poster works. The wording doesn't.
MiniMax H3 Image is now on NanoGPT. Upload an image and describe the edit, or start from a prompt. Up to 2K, with 15 aspect ratios. $0.03 per generated image. $0.04 per edit.
nano-gpt.com/studio/image?mo…
3
499
Gemini 4 Carbon is now available on NanoGPT!
nano-gpt.com/conversation?mo…
10
867
Claude Fable 5.2 is now on NanoGPT. Anthropic reports 49.1% on FrontierCode 2.0 at Xhigh effort and 69.9% on FrontierSWE v2.
It supports text, images and PDFs, tool calling, structured output, a 1M-token context and up to 128K output tokens.
Anthropic retains prompts and outputs for 30 days; zero data retention is unavailable for this model on NanoGPT.
nano-gpt.com/conversation?mo…
1
2
407














