NaN
Shared dedicated GPUs for running open models, with an OpenAI-compatible API
About NaN
NaN is for developers who want to run open models seriously, providing inference through shared dedicated GPUs. Members can access the shared cluster and community-selected open models, and use a personal API key compatible with OpenAI. The member plan includes DeepSeek V4-Flash inference capacity; the premium plan provides GLM 5.3, 1M context, and concurrent requests, making it suitable for independent developers, makers, agents, and launched side projects. The service uses a zero-log design, with processing in the EU.
Best for
Best use cases
Run open models
Add inference capabilities to agents and side projects
Call models through an OpenAI-compatible API
Use vision and reasoning models
Features
Shared GPU cluster
Run open models through shared dedicated GPUs and access community-selected models.
OpenAI-compatible API
Members receive a personal API key compatible with OpenAI.
DeepSeek V4-Flash
The nan_member plan includes this model and 3B tokens per month.
GLM 5.3 Premium
The premium plan provides 1M context and 5 concurrent requests.
Pros
- Shared cluster supports open models with no token caps
- Personal API key compatible with OpenAI
- Zero logs; processing takes place in the EU
- Open models selected by the community
Cons
- GLM 5.3 premium is limited to 3,000M tokens per billing period and 400M tokens per rolling 4-hour window
- The service is for inference, not training or fine-tuning
Pricing Plans
nan_member inference included
- Access to the shared cluster
- Open models with no token caps
- DeepSeek V4-Flash, 3B tokens/month
- Personal API key compatible with OpenAI
- Private Discord channels for members
nan_member · glm 5.3 premium
- GLM 5.3
- 3,000M token allowance per billing period
- 400M tokens per rolling 4h window
- 1M context · 5 concurrent requests
- Everything in nan_member
* Pricing information is for reference only. Please refer to the latest announcement on NaN's official website.