Claude Code sends 33k tokens before reading the prompt; OpenCode sends 7k
An analysis reveals that Claude Code sends a considerably higher volume of tokens compared to OpenCode before it begins to process a user's prompt. Before reading the prompt, Claude Code transmits 33,000 tokens. In contrast, OpenCode sends 7,000 tokens in the same pre-processing phase. This means Claude Code sends 4.7 times more tokens than OpenCode before the user's input is engaged. This increased token overhead for Claude Code occurs prior to the model's interaction with the actual prompt, which may influence the overall resource consumption and operational costs for applications developed using this model.
Developers using Claude Code may incur higher token costs due to the model sending significantly more tokens before processing prompts.