AI tokenomics
Independent testing by Tolly
Provide the following detail to gain access to the full report
Independent testing by Tolly shows that F5 BIG-IP Next for Kubernetes (BNK) significantly improves AI inference performance compared with traditional open-source load balancing.
Results
No changes to models
No changes to inference frameworks
No application rewrites
40%+ MORE AI TOKENS
Higher token throughput
More AI output from the infrastructure you already have
60%+ FASTER FIRST TOKEN
AI responses start faster
Better user experience
30%+ LOWER LATENCY
Faster end-to-end inference responses
Why it Matters:
AI models generate value one token at a time.
The infrastructure delivering those tokens determines how fast, efficiently, and reliably your AI systems operate.
Independent testing shows how optimizing the AI delivery layer can significantly improve inference performance.