Channelwise FP8 weights, dynamic per-token activations
2 published packages
- Measured models
- AREX-Turbo and Qwythos-9B-Claude-Mythos-5-1M
- Accuracy recovery
- 99.35% to 107.14%
Optimization
Measured optimization techniques for open model inference
The published catalog currently carries 2 measured techniques across 3 published packages.
Channelwise FP8, measured selective-layer recipe
1 published package
- Measured models
- Qwen3.6 27B
- Accuracy recovery
- 99.87%
If you need custom optimization for a specific model, describe what you need
End-to-end encryption
Isolated GPU infrastructure
No training on your data
SOC 2 Type II
RunInfraby RightNow© 2026 RunInfra. All rights reserved.
