DeepSeek / DeepSeek V4 Pro 0813
Mixed FP4 / FP8 quantization for DeepSeek V4 Pro 0813
The published profile is indicative; engineering measurements are completed for the selected model and hardware.
Before → After
Base releaseVerified checkpoint
1650.498B
1650.498B
QuantizationMixed FP4 / FP8
Quality retention target
Quality retention target
Production artifactArtifact
Lineage and benchmark report
Lineage and benchmark report
GPU targeting
Weights and KV cacheEstimated until measured during engineering
Estimated VRAMMeasured and certified results are delivered with the project.
Runtime packagingtransformers
Transparent pricing
Not currently orderable