Frequently Asked Questions

ChatGPT processes your documents on shared infrastructure, and your data may be used for training. ArcticAI spins up a dedicated GPU instance for each chamber. Your documents are processed only on that instance, are never shared with other customers, and are never used for model training. When you destroy the chamber, the instance and its data are deleted.

Your documents are uploaded over HTTPS to ArcticAI, which passes them straight to your dedicated GPU instance without keeping a copy. They are processed on that instance using open-source AI models. Our database holds only metadata (chamber status, billing, file names in your audit log) and your chat history only if you choose to save it.

When you destroy a chamber, the dedicated GPU instance is terminated and deleted together with its storage — uploaded documents, AI model memory and generated outputs. Any chat history you chose to save is deleted from our database at the same time. This process is irreversible by design.

We use open-source models served via vLLM. The Flash tier runs Qwen3-4B Instruct and Turbo runs Qwen3.5-35B-A3B (an AWQ-quantized Mixture-of-Experts model). Ultra, which will run GLM-5 744B, is coming soon. All model weights are fully auditable and run entirely on your dedicated GPU instance — no data is sent to any third-party API.

ArcticAI is not HIPAA certified and does not sign business associate agreements. Each chamber runs on its own dedicated GPU instance that no other customer uses, and the instance is deleted when you destroy the chamber. Check with your compliance team before handling regulated health data.

ArcticAI uses simple pay-as-you-go pricing. GPU time is $0.99/hr for Flash (Qwen3-4B) and $4.99/hr for Turbo (Qwen3.5-35B-A3B). Ultra (GLM-5 744B) is coming soon and can’t be ordered yet. You only pay for the minutes your chamber is running. Add credit to your balance and start analyzing — no subscriptions or commitments required.

Boot time depends on the GPU host. Flash chambers usually take about 5 minutes and Turbo about 8 minutes, because Turbo downloads a larger 25 GB model at startup. These are estimates, not guarantees. Boots on a host you have used before are usually faster.

ArcticAI supports PDF, DOCX, TXT, and other common document formats. Files are processed on your GPU instance using OCR and text extraction. The maximum file size is 50 MB per file on every account.

Our servers pass your documents to your chamber but do not store or inspect them. We keep metadata such as chamber status, billing and the file names in your audit log, and your chat history only if you turn saving on. The documents themselves live only on your dedicated GPU instance.

Yes. When you have a document selected, type an instruction starting with 'edit', 'update', 'change', or similar keywords. The AI will modify your document according to your instructions and show you a diff of the changes. You can review and accept or reject the edits before saving.

We provision GPUs from a global network of data centers. Flash tier runs on a single GPU with at least 16GB of memory. Turbo uses L40S 45GB GPUs for the larger model. Ultra (coming soon) will require 8x H200 141GB GPUs for the massive 744B parameter model. Instances are provisioned from data centers in North America, Europe, or Australia with high-bandwidth connections — see our Privacy Policy for details on data location.

Auto-destruct is a safety feature that automatically destroys your chamber after a set period (1, 4, 12, 24, or 48 hours). This ensures that even if you forget to manually destroy a chamber, your sensitive data won't persist indefinitely. Auto-destruct also stops billing, so you won't be charged for idle chambers.

Uploads from your browser to ArcticAI use HTTPS. On the instance, the storage is ephemeral: when the chamber is destroyed, the instance and its disk are deleted. We never store your documents on our own servers.

Closed-source models like GPT and Claude require sending your data to third-party AI providers (OpenAI, Anthropic). With open-source models, the AI runs entirely on your dedicated GPU — your documents are never sent to an external AI service. The model weights are also public, so you can check exactly what is processing your documents.

Unused credits are refundable within 14 days of purchase. GPU time that has already been consumed is non-refundable, as the compute resources are provisioned and billed in real time. If you experience a technical issue that causes unexpected charges, contact support and we'll review your case.

Currently, chambers are single-user. Each chamber is tied to one account, and shared or multi-user access isn't supported yet. Reach out if team access is something you need — it helps us prioritize the roadmap.