Claude Code uses approximately 33,000 tokens of system prompt, tool schemas, and injected scaffolding before processing a prompt, compared to OpenCode's 7,000 tokens. This results in Claude Code sending 4.7 times more tokens than OpenCode before reading user input.

Meta has capped internal AI token spending as costs are projected to reach billions of dollars by 2026. Employees consumed 73.7 trillion tokens in about 30 days, prompting CTO Andrew Bosworth to emphasize that high token usage does not equate to meaningful progress.