Compare

Claude Opus 4.7 vs GPT-5 Mini

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

Verdict

Claude Opus 4.7 leads on LMSYS ELO (1412 vs 1288). GPT-5 Mini is ~34.3× cheaper on a 3:1 input:output blend.

Anthropic

Anthropic’s frontier reasoning model. Leads SWE-Bench and holds top ranks on long-context coding and agentic workflows.

Best-in-class coding agent behaviorStrong instruction following1M-token context with high recallHigh reliability under tool use

OpenAI

GPT-5’s smaller, cheaper sibling. Optimized for chat and lightweight agent tasks.

Very low latencyCheap enough for mass consumer apps

Pricing

Input / 1M$15.00$0.50

Output / 1M$75.00$2.00

Context1.0M256K

Max output64K16K

LicenseProprietaryProprietary

Released2026-03-122026-01-21

Benchmarks

LMSYS ELO1412.01288.0

MMLU Pro92.878.5

HumanEval94.485.5

SWE-Bench72.1—

MATH89.379.8

GPQA66.4—

MMMU79.2—

IFEval92.185.4

Per-task cost

Summarize a 1-hour meeting transcript$0.263$0.0085

Review a 500-line pull request$0.270$0.0080

Answer a customer support ticket$0.105$0.0031

Extract structured data from a resume$0.150$0.0045

Debug a stack trace with context$0.240$0.0070

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

Add up to four models, tweak tasks live.