Ollama benchmark raw data 2025H1

Ollama eval rate of benchmark raw data collected from January, 2025 to June, 2025.
PII data won't be included.

Around 850+ different combinations of hardware specs.

More than 9000 records of LLM models, hardware spec and eval rate (throughput).

It helps buyers decide which system (cpu, gpu) should be bought.

Who Benefits?

• IT Teams: Streamline your internal LLM deployment process with actionable data.
• Decision Makers: Make confident hardware purchasing decisions grounded in real-world metrics.
• Developers: Accelerate model optimization with precise, platform-specific benchmarks.