YOU
your device
›
‹ A7F3A7F3·A7F3
Open-weight GLM-5.3 and Qwen-3.8. No logs. History and images stay on your device. OpenAI-compatible API.
Your prompts are computed, not collected.
$ ask --private
you › Write a streaming API client in Python
glm › Here is a client that prints tokens as they arrive.
nothing above was logged — it lives only in your browser
open the live terminalPrompts and responses travel over TLS and are processed in memory for inference. History and images stay on your device.
your device
›
‹ A7F3A7F3·A7F3
our hardware
Inference runs on GPUs we own and operate. We run the models ourselves rather than reselling a third-party inference API.
History and images stay on your device. No server-side conversation storage or training on your data. Billing metadata never contains prompts or responses.
Paid chat and API have no general-purpose content filtering. Trial chat has a narrow illegal-content filter. Lawful use under our AUP is required.
Pay-as-you-go credit packs, or a monthly plan for regular use — either way you're buying compute, never locked in.
20 free credits to try · expire after 14 days without a top-up · no logs
Access to all models · 32K context · No API access · No web search
Loading current offers…
Active GM staking earns prefunded rewards and a time-weighted share of compute from the hardware we operate. Claim daily CMP when an epoch budget is available.
For creators & influencers
Get paid a percentage when your audience buys AI credits. We review your content and audience size to offer you a personal revenue-share rate.