
1 / 3
【Overview】 NaN is a builders' community that pools dedicated GPUs to run open AI models behind an OpenAI-compatible API. Instead of paying per token, members pay a flat monthly subscription and get an endpoint at https://api.nan.builders/v1 that works with any tool accepting a base URL and API key. The cluster hosts 13 open models 24/7, with the model lineup chosen by community vote every quarter. The service targets indie hackers, makers, and builders running agents and side projects in production who want to use open models seriously without standing up their own infrastructure. It handles the GPU cluster, the API surface, and isolation so you can focus on your application. Beyond inference, NaN provides isolated Linux workspaces with coding agents preinstalled, autonomous agent runs, and automations triggered by webhooks or schedules. All inference is processed in the European Union with zero logging of prompts or responses and no training on your code. 【Key features】 • OpenAI-compatible inference API: Serves open LLMs, embedding, rerank, TTS/STT and image models behind an OpenAI-compatible API at https://api.nan.builders/v1, so any tool that accepts a base URL and API key works. • Shared GPU inference cluster: Runs 13 open models 24/7 on shared dedicated GPUs, chosen by community vote every quarter. • Workspaces: Provides a dedicated, isolated Linux cloud machine reachable over SSH or a browser terminal, with coding agents such as Pi, Hermes, gentle-shell and Herdr preinstalled. • Workspace backups add-on: Takes a daily snapshot of the workspace disk and lets you restore it yourself from the Backups tab. • Agent runs: Gives an agent installed in your workspace (Pi, Pi + Gentle Shell or Hermes) a task to work on autonomously, with a live log and changes landing on a branch in git repos. • Automations: Saves an agent task and starts it from a webhook, a schedule or your scripts, with approvals you choose. 【Use cases】 1. Connect coding agents (Claude Code, Codex, Cursor, Cline, OpenCode, Zed, VS Code Copilot) to a shared inference cluster 2. Run agents on tasks inside your own cloud workspace 3. Start agent tasks from webhooks or schedules as automations 4. Build RAG and semantic search apps (e.g. bilingual medical document system) 5. Generate images and speech with cluster image and TTS/STT models 【Pricing】 nan_member: 70€/month — Access to the shared cluster. No token counter on the cluster models., DeepSeek V4-Flash (frontier 305B, 1M context, vision, reasoning). 3B tokens/month., Open models chosen by the community, Personal API key compatible with OpenAI glm 5.3 premium: 200€/month — GLM 5.3, frontier open model with reasoning, 3,000M token allowance per billing period, 400M tokens per rolling 4h window: the limit a coding agent reaches first, 1M context · 10 concurrent requests nan_community: 14,99€/month — Discord access, Member-only channels, Live discussions, builds, AMAs, Events, workshops and hackathons Micro workspace: 4,99€/month — 1 vCPU · 2 GiB RAM · 10 GiB disk, 0,99€ /mo optional daily backups Nano workspace: 8,99€/month — 2 vCPU · 4 GiB RAM · 20 GiB disk, 1,49€ /mo optional daily backups 【Pros and cons】 Pros: - Flat-rate pricing removes per-token anxiety for builders running agents and side projects continuously. - Community-driven model selection keeps the cluster aligned with what members actually need. - EU-based zero-log inference appeals to developers handling sensitive code or data. Cons: - The Community plan cannot create an API key; inference requires nan_member or GLM 5.3 premium. - Frontier models have token allowances per member; only certain cluster models are unmetered. - Workspaces do not include inference and have no on-demand backups without the add-on. - Daily backups are kept for 7 days with up to 3 restores per workspace in any 24 hours. 【FAQ】 Q: Can I use my existing OpenAI-compatible tools with NaN? A: Yes, the API is OpenAI-compatible and available at https://api.nan.builders/v1, so any tool that accepts a base URL and API key should work. Q: What is the difference between the nan_member and glm 5.3 premium plans? A: nan_member costs 70€/month and gives access to the shared cluster with a 3B tokens/month allowance on DeepSeek V4-Flash, while glm 5.3 premium costs 200€/month and provides a 3,000M token allowance per billing period for GLM 5.3 with 10 concurrent requests. Q: Is NaN suitable for training or fine-tuning models? A: No, NaN is designed for inference and running agents, not for training or fine-tuning. It is best for indie hackers, makers, and builders running agents and side projects in production.
nan_member · 70€/month (Access to the shared cluster. No token counter on the cluster models., DeepSeek V4-Flash (frontier 305B, 1M context, vision, reasoning). 3B tokens/month., Open models chosen by the community) glm 5.3 premium · 200€/month (GLM 5.3, frontier open model with reasoning, 3,000M token allowance per billing period, 400M tokens per rolling 4h window: the limit a coding agent reaches first) nan_community · 14,99€/month (Discord access, Member-only channels, Live discussions, build
WPS AI is an AI assistant built into WPS Office that helps users write documents, generate presentation outlines and…
Baidu Comate is an AI coding assistant built by Baidu on its ERNIE model. It goes beyond simple autocomplete: it…
UniConverter is an AI media conversion and enhancement tool for video, audio, and images. It combines everyday…
6 similar tools to consider instead of NaN