Compare

GPT-5 Turbo vs o4

OpenAI vs OpenAI. Specs, benchmarks, and real per-task cost — all in one page.

Verdict

GPT-5 Turbo is ~2.5× cheaper on a 3:1 input:output blend.

OpenAI

OpenAI’s flagship unified model. Handles text, vision, and audio natively. The generalist benchmark champion.

Native multimodalStrong math and scienceHuge ecosystem

OpenAI

Deep-reasoning model. Spends more tokens thinking to crush math, science, and hard coding problems.

State-of-the-art math and scienceExtremely high-ceiling reasoning

Pricing

Input / 1M$6.00$15.00

Output / 1M$24.00$60.00

Context400K200K

Max output32K100K

LicenseProprietaryProprietary

Released2026-01-212026-02-18

Benchmarks

LMSYS ELO1398.0—

MMLU Pro91.293.8

HumanEval93.092.3

SWE-Bench65.868.0

MATH91.596.2

GPQA68.182.5

MMMU78.8—

IFEval88.9—

Per-task cost

Summarize a 1-hour meeting transcript$0.102$0.255

Review a 500-line pull request$0.096$0.240

Answer a customer support ticket$0.037$0.093

Extract structured data from a resume$0.054$0.135

Debug a stack trace with context$0.084$0.210

Per-call cost using published token counts for each task. Real-world prompts vary.

Add up to four models, tweak tasks live.