Open-model inference, served from Canada
InferStack runs open-weight language and vision models on dedicated GPUs and offers them through an OpenAI-compatible API, available via OpenRouter.
Models
| Model | Inputs | Context |
|---|---|---|
| Qwen3-VL-8B-Instruct | text, image | 32K tokens |
How we handle your data
- Prompts and completions are processed in memory and not stored.
- We never train on customer data.
- We keep only billing metadata (request ID, model, token counts, timings, status).
Details: Privacy policy.
API
Base URL https://api.inferstack.ca/v1. Access is provided through OpenRouter.
Contact
hello@inferstack.ca · Moncton, New Brunswick, Canada