invokeable()

Claude Opus 5 vs DeepSeek V4 Pro 0813

A custom comparison card. Numbers link back to their entries; data as of 2026-08-19. Build your own →

model
Claude Opus 5
Anthropic
model
DeepSeek V4 Pro 0813
DeepSeek
Context1M1M
Input $/M
Output $/M
Knowledge cutoff2026-05
Available viaAnthropic APIDeepSeek API
SWE-bench Verified (resolved)96
SWE-bench Pro79.2%
SWE-bench Multilingual89.5%
SWE-bench Multimodal59.4%
DeepSWE68.8%
FrontierCode (mean@5)53.4