InferBench
InferBench is an open benchmark platform for open-source AI models on real hardware.
Measure throughput, latency, energy, and serving economics, then compare results on a public leaderboard.
JavaScript is required for the full interactive experience. Key public pages:
- Leaderboard — compare open models and hardware from real runs
- Benchmark — run the InferBench CLI against your hardware
- Docs — guides, CLI reference, and methodology
- Pricing — synced model prices for reported benchmarks
- Calculator — estimate serving yield from benchmarked hardware
- GitHub repository