Lower eligible input spend
Focus on measurable input-token reduction rather than a universal savings promise.
Token Optimisation
Spend less on eligible AI input traffic.
Measure the result on your own workload before committing production traffic.
100,000 free processed input tokens + 50 local test runs after email verification · Results vary by workload
Measure first
Prove the value on your own traffic.
Use the free testing allowance before a production decision.
Validate before rollout
Compare results on representative requests before production adoption.
Independent product access
Use Token Optimisation without purchasing Forge or Compute.
NVIDIA CUDA acceleration
Available for supported high-volume retrieval workloads, with automatic CPU fallback when a compatible NVIDIA CUDA runtime is not available.
Packages
Token Optimisation pricing
Prepaid processed-input-token packages.