Compare
DeepSeek V4 vs GPT-5 Mini
DeepSeek vs OpenAI. Specs, benchmarks, and real per-task cost — all in one page.
Verdict
DeepSeek V4 leads on LMSYS ELO (1334 vs 1288). DeepSeek V4 is ~5.0× cheaper on a 3:1 input:output blend. DeepSeek V4 is open-weights, the other is proprietary.
DeepSeek
DeepSeek V4The shock-the-market moment of 2026. Frontier coding quality at 1/50th the price of Opus.
Unbeatable $/qualityOpen weightsGreat at code
OpenAI
GPT-5 MiniGPT-5’s smaller, cheaper sibling. Optimized for chat and lightweight agent tasks.
Very low latencyCheap enough for mass consumer apps
Pricing
Input / 1M$0.14$0.50
Output / 1M$0.28$2.00
Context128K256K
Max output16K16K
LicenseOpen weightsProprietary
Released2026-02-202026-01-21
Benchmarks
LMSYS ELO1334.01288.0
MMLU Pro85.478.5
HumanEval91.585.5
SWE-Bench54.1—
MATH88.779.8
IFEval85.885.4
Per-task cost
Summarize a 1-hour meeting transcript$0.0022$0.0085
Review a 500-line pull request$0.0017$0.0080
Answer a customer support ticket$0.0006$0.0031
Extract structured data from a resume$0.0010$0.0045
Debug a stack trace with context$0.0014$0.0070
Per-call cost using published token counts for each task. Real-world prompts vary.