BENCHMARKFirst benchmarks recorded: MiniMax M3 → Cortecs → ?3hBENCHMARKFirst benchmarks recorded: Kimi K3 → Cortecs → ?4hBENCHMARKFirst benchmarks recorded: DeepSeek V4 Pro → Cortecs → ?4hBENCHMARKFirst benchmarks recorded: GLM-4.7 → Cortecs → ?4hBENCHMARKFirst benchmarks recorded: GLM-5.2 → Cortecs → ?4h
MARKET INTELLIGENCE

The Price–Performance Frontier of AI Inference

Which providers are actually efficient — and which are dominated.

Aug 15, 2026103,921 benchmark runs13 providers8 modelsMethodology v0.10-devDEV DATA
Key findings
  • 3 of 13 providers are Pareto-dominant on score × output price under the General profile.
  • 77% of the market is dominated: a strictly better and cheaper alternative exists.
  • The frontier is workload-sensitive — providers enter and leave it when the profile shifts to Agents or Realtime.

The frontier

Ranking position and value position are different questions. The scatter below marks the undominated set: Nebius, DeepInfra, Nextbit. Everything else is inside the frontier.

78838894$0.31$0.62$0.92$1.23Output price $/M →Score ↑Nebius
Score vs output price · Pareto set highlighted · DEV DATA
What this means for engineers
  • If a provider is dominated under your workload profile, the burden of proof is on the other axes: residency, capabilities, support.
  • Price cuts move the frontier — watch the ticker's PRICE events; they reshape this chart within a day.
Dataset
103,921 runs30-day rolling window · Aug 2026snapshot 2026-08-15-v1Methodology v0.10-devDEV DATA

Published figures are pinned to this snapshot and are not recalculated retrospectively. View the current benchmark →

Cite this researchInferenceBench (2026). The Price–Performance Frontier of AI Inference. InferenceBench Research.
← All research
Related research
MARKET REPORT

The State of AI Inference — August 2026

The flagship snapshot of the inference market: the global provider ranking, the price–performance frontier, and the growing spread between infrastructures serving identical models.

Aug 15, 2026MARKET REPORT

The European AI Inference Report

The deepest public look at European inference: EU-owned providers versus Europe-hosted endpoints, the residency premium, and where the region genuinely competes.

Aug 15, 2026ENGINEERING STUDY

Same Model, Different Infrastructure: How Large Is the Gap?

The core InferenceBench thesis, measured: for the same model, TTFT varies by multiples, throughput by an order of magnitude, and tool reliability by whole points.

Aug 15, 2026