This page may contain stale information. Last updated: 2026-08-14

Definition

Ultrafast is an OpenAI API service tier for gpt-5-6-sol that serves the same model at substantially higher output token rates using cerebras wafer-scale-engine hardware.

Key Points

  • Launch: Aug 13, 2026 — limited preview for select API customers
  • Claims: up to 750 output tokens/sec; up to 14× Standard
  • Vendor: same intelligence as Standard; GDP-Val 5.6× end-to-end with no quality loss
  • Competitive framing vs claude-opus-4-8 Fast and claude-fable-5 (Artificial Analysis–cited)
  • Target workloads: incident response, support, financial analysis, e-commerce

Speed and "no quality loss" claims are vendor / Artificial Analysis cited — treat as claimed until independently verified. Pricing undisclosed.

Sources