Grok 4.6 is now available on NanoGPT. Cursor reports an Artificial Analysis Intelligence Index score of 61, plus 1753 on GDPval-AA v2, 69.9% on CursorBench v3.2, and 57.5% on APEX-Agents. Try it:

NanoGPT

NanoGPT

Access the newest AI models including ChatGPT, Claude, Gemini, Deepseek, and image/video models. Pay-as-you-go or subscribe. Local history by default.

DeepSeek V4 Pro 0813 is now available on NanoGPT, with an optional Thinking variant. DeepSeek reports major gains over the preview, with scores above Opus 4.8 on Terminal Bench 2.1, Cybergym, DeepSWE, and AutomationBench. Included in the subscription. Prompts may be logged and used for training. Try it:

NanoGPT

NanoGPT

Access the newest AI models including ChatGPT, Claude, Gemini, Deepseek, and image/video models. Pay-as-you-go or subscribe. Local history by default.

Wan 2.5 Fast and Wan 2.5 Extend Fast are now available on NanoGPT. Generate video with synchronized audio from a prompt, or continue an existing clip while preserving its motion and sound. Both support optional custom audio. Try them:

Media | NanoGPT

Generate and manage images, videos, and audio with NanoGPT.

Wan 2.2 Animate 2 is now available on NanoGPT. Transfer motion from a driver video to a character in a reference image while preserving their identity. Create performances, demonstrations, and motion-driven stories up to 120 seconds long. Try it:

Media | NanoGPT

Generate and manage images, videos, and audio with NanoGPT.

Kimi K3 is now available in NanoGPT Private Mode. Use its long-context reasoning, coding, and agent capabilities with browser-to-enclave encryption. NanoGPT can route and bill the request without being able to read your prompt or response. Try it:

NanoGPT

NanoGPT

Access the newest AI models including ChatGPT, Claude, Gemini, Deepseek, and image/video models. Pay-as-you-go or subscribe. Local history by default.

Seedance 2.5 is now available on NanoGPT, alongside Turbo and Spicy variants. Create cinematic videos from text, images, references, or existing clips, with native audio, editing, and extension. Try it:

Media | NanoGPT

Generate and manage images, videos, and audio with NanoGPT.

Building agents that delegate work to other agents? Our new A2A guide shows how to use NanoGPT's API inside an Agent2Agent app, with Python setup, Agent Cards, authentication, architecture, and a production checklist. Read it:

Build A2A Agents with NanoGPT

Connect an A2A agent to NanoGPT for model inference without exposing your NanoGPT API key in its Agent Card.

Qwen Image 3 Pro is now available on NanoGPT. Generate detailed 1K or 2K images, render strong typography, or make precise edits using up to three reference images. Create or edit:

Media | NanoGPT

Generate and manage images, videos, and audio with NanoGPT.

Muse Spark 1.2 is now available on NanoGPT. Built for end-to-end coding across large repositories, it can plan changes, write code, validate results, and work as a lead agent or parallel subagent. Try it:

NanoGPT

NanoGPT

Access the newest AI models including ChatGPT, Claude, Gemini, Deepseek, and image/video models. Pay-as-you-go or subscribe. Local history by default.

FLUX.3 is now available on NanoGPT. Generate videos up to 20 seconds with native audio and multilingual dialogue from a prompt, start image, start and end frames, keyframes, or an existing clip. Try it:

Media | NanoGPT

Generate and manage images, videos, and audio with NanoGPT.

Pokee-Isaac 28B is now available on NanoGPT. Built for long-context agentic work, it supports a 10M-token context window and is designed to run on a single GPU. Pokee reports 93.3% on RULER at 10M tokens and up to 137K tokens/s prefill. Try it:

NanoGPT

NanoGPT

Access the newest AI models including ChatGPT, Claude, Gemini, Deepseek, and image/video models. Pay-as-you-go or subscribe. Local history by default.

Prompt caching is easier to measure on NanoGPT. The Usage API now reports cache creation and cache read input tokens, with coverage counts for each bucket. CSV and JSON usage exports include both cache token fields too. Inspect usage:

Join NanoGPT

Create your NanoGPT account and get access to leading AI models for chat, coding, images, and video.

Failed API request? NanoGPT's Usage page can now tell you why. See the HTTP status, likely cause, retry timing, duration, API key, and request ID for Chat Completions errors, including failures without a normal usage entry. Request and response bodies aren't stored with these diagnostics. Inspect your API usage and errors:

Join NanoGPT

Create your NanoGPT account and get access to leading AI models for chat, coding, images, and video.

Qwen Audio 3.0 TTS Flash is now available on NanoGPT. This fast speech model supports 10 languages and a broad voice library for narration, dialogue, localization, and multilingual audio workflows. Generate speech:

Media | NanoGPT

Generate and manage images, videos, and audio with NanoGPT.

MiniMax H3 is now available on NanoGPT. Generate 2K video at 24 FPS with native stereo audio from text or an image. Create first-to-last-frame transitions, or guide subjects, motion, and sound with image, video, and audio references. Create with H3:

Media | NanoGPT

Generate and manage images, videos, and audio with NanoGPT.

New in NanoGPT: image annotations ✏️ Open an uploaded image, tap Draw, and circle exactly what you want the model to focus on. Every mark is labeled automatically for easy reference. No more “the thing in the top-right.” Just show the model what you mean.

🔒Passkey-protect your conversations! Your most sensitive chats can now hide behind your passkey. One tap on the fingerprint icon and that chat is blurred in your lists, hidden from search, and locked on screen until you verify with Face ID, Touch ID or a security key. Find the option in the conversation header.

Tron USDT deposits are now direct on NanoGPT and no longer needs a third-party checkout. ⚡ Faster deposits. 💸 Lower fees, less friction. 🔒 Better privacy. Another step toward bringing our crypto payments fully in-house! #TRON #USDT #TRONics

Fugu Ultra is now available! It coordinates a pool of expert agents to tackle hard problems where answer quality matters most. Fugu Ultra scores 73.7 on SWE Bench Pro, 82.1 on TerminalBench 2.1, 93.2 on LiveCodeBench, 90.8 on LiveCodeBench Pro, and 50.0 on Humanity's Last Exam, putting it ahead of listed public frontier baselines on several coding, agentic, scientific, and reasoning tests. Try it for yourself! #AI #Agents #Coding #LLM #Reasoning #Fugu #NanoGPT

NanoGPT now renders SVGs directly in chat. You can preview the vector, switch back to code, and export the result as PNG. Useful for logos, icons, diagrams, and yes, small penguins! 🐧