LLM Model Comparisons

Compare model families across provider-hosted endpoints and inspect how serving provider, configuration, benchmark, and prompt mode affect results.

Model Families

A model family is the underlying model identity, such as a GPT, Claude, Gemini, Llama, or Mistral family member.

Hosting Providers

The same model family can behave differently across hosted endpoints because providers may vary quantization, infrastructure, rate limits, or serving configuration.

Model Families With Public Runs

  • Llama 3.3 70B: 27 public runs across 2 endpoints.
  • Qwen 3 235B A22B Instruct: 9 public runs across 2 endpoints.

Benchscope is a JavaScript app. If the interactive interface does not load, enable JavaScript or use the links above for the main public sections.