token.app โ€บ Compare โ€บ GPT-5 vs DeepSeek V4 Pro

GPT-5 vs DeepSeek V4 Pro

DeepSeek V4 Pro leads on 4 of 6 shared benchmarks. DeepSeek V4 Pro is 6.3ร— cheaper per token on a blended 3:1 basis. DeepSeek V4 Pro accepts the larger context window (1.0M).

Head to head
Blended price uses a 3:1 input:output mix
GPT-5DeepSeek V4 Pro
ProviderOpenAIDeepSeek
Input $/1M$1.25$0.435
Output $/1M$10.00$0.870
Blended $/1M$3.44$0.544
Context window400K1.0M
Max output128K384K
ReleasedAug 2025Apr 2026
LicenceProprietaryOpen weights
GPQA Diamond 86.2% 90.9%
SWE-bench Verified 73.6% 77.6%
FrontierMath T1โ€“3 55.4% 45.3%
FrontierMath T4 22.0% 2.4%
SimpleQA Verified 50.6% 57.0%
OTIS Mock AIME 91.4% 96.7%
Benchmark scores independently run and published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0. Only benchmarks with a score for BOTH models are compared.
Choose GPT-5 ifโ€ฆ
  • Higher FrontierMath T1โ€“3 (55.4% vs 45.3%)
  • Higher FrontierMath T4 (22.0% vs 2.4%)
Choose DeepSeek V4 Pro ifโ€ฆ
  • Costs less per token โ€” $0.544 vs $3.44 blended
  • Higher GPQA Diamond (90.9% vs 86.2%)
  • Higher SWE-bench Verified (77.6% vs 73.6%)
  • Higher SimpleQA Verified (57.0% vs 50.6%)
  • Higher OTIS Mock AIME (96.7% vs 91.4%)
  • Larger context window (1.0M)
  • Open weights โ€” self-hostable
  • Newer model (released Apr 2026)
FAQ
Which is better, GPT-5 or DeepSeek V4 Pro?
Across the 6 benchmarks both models have independently-run scores for, GPT-5 leads on 2 and DeepSeek V4 Pro leads on 4. Benchmarks are one input โ€” price and context window are on this page too.
Is GPT-5 cheaper than DeepSeek V4 Pro?
DeepSeek V4 Pro is cheaper: $0.544 versus $3.44 per million tokens blended at 3:1 input:output. Input alone: $1.25 vs $0.435. Output alone: $10.00 vs $0.870.
What context windows do GPT-5 and DeepSeek V4 Pro have?
GPT-5: 400,000 tokens. DeepSeek V4 Pro: 1,048,576 tokens.
Who makes GPT-5 and DeepSeek V4 Pro?
GPT-5 is made by OpenAI. DeepSeek V4 Pro is made by DeepSeek.
Related comparisons