A free 30-minute pass to GLM-5.3-Flash
Spark Queue hands out free API passes, one guest at a time, for one model served on two linked NVIDIA DGX Spark units.
- GLM-5.3-Flash: an OpenAI- and Anthropic-compatible API.
- A 1M-token context (1,048,576 tokens).
- Speed: one stream about 40 tokens/s for prose and 60 for code (measured 2026-11-05).
- Works with OpenCode, pi and Claude Code.
The queue now
0 people wait. No pass is in progress.
The queue is paused. Offers wait until it resumes.
The pair is offline. Offers wait until it is back.