Clarifai has launched a new reasoning engine designed to optimise AI model performance and reduce costs. The company claims the engine can double the speed of AI models and reduce expenses by 40%. The engine uses optimised kernels and techniques that dynamically adapt to workloads, improving generation speed without sacrificing accuracy.
The reasoning engine is designed to accelerate reasoning models and automation tasks across industries. It targets the inference stage, where trained AI models execute tasks. Clarifai's engine has demonstrated industry-leading results for throughput and latency on GPUs, outperforming some specialised non-GPU accelerators. The platform's adaptive performance learns from workload behaviour, improving speed over time.
