Models
How Bloodhound selects and meters its investigation engine.
The public API exposes Bloodhound capabilities rather than provider-specific model selection. The current production investigator uses deepseek-v4-flash with high investigation depth.
Model selection is managed by Bloodhound so prompts, tools, metering, and safety limits remain consistent. A future model selector will be introduced only with a stable compatibility contract.
Metered usage
Input, cache-hit input, cache-miss input, and output are recorded separately. See pricing for published customer rates.
Next steps
Read investigations to understand how the engine plans and verifies a case.