DevToolsReview’s widely referenced Aider pricing guide is currently misleading developers about their actual API spend. The page’s entire cost calculator relies on Claude 3.7 Sonnet rates, a model generation that Anthropic has effectively superseded. By anchoring every worked example, from daily spend ranges to the head-to-head against Cursor Pro, to the outdated $3.00/$15.00 per million token price, the guide overstates real-world costs by roughly a third for any user running today's default Sonnet model.

The Model Gap Is Two Generations Wide

Anthropic shipped Claude Sonnet 5.5 on September 28, 2026, establishing it as the new default Sonnet model with a 1M-token context window. Crucially, this launch included a repricing to $2 per million input tokens and $10 per million output tokens. The Aider guide, however, still explicitly states in its FAQ that "Claude 3.7 Sonnet is currently the best overall performer for coding tasks," a claim that ignores the existence of the newer, cheaper, and more capable model. This isn't just a stale price on a current model; it is a fundamental misalignment with the current state of Anthropic's offering.

Real-World Cost Implications

The financial discrepancy is not negligible. For a developer using Aider with today's default configuration, the guide's estimate of $30-$70 per month for moderate use is inflated; the actual cost falls closer to $20-$47 per month. Similarly, the per-message input cost for a 10-file refactor, listed as $0.15-$0.30, should be roughly $0.10-$0.20. This error systematically makes Aider appear more expensive than it is, potentially steering users toward flat-rate subscriptions like Cursor Pro when the BYOK (Bring Your Own Key) model is actually more economical than the guide suggests.

The Fragility of BYOK Pricing Pages

This incident highlights a structural weakness in documentation for BYOK tools. Unlike flat-rate services where pricing changes are internal and predictable, Aider's cost profile is entirely dependent on third-party API rate cards. When Anthropic repriced or launched new models, nothing in Aider's documentation pipeline triggered an update. The guide's authors have confirmed via internal freshness tracking, sourced to Anthropic's Claude Code release notes (v2.1.284), that the correction is pending. Until the full page is revised, developers should treat the Claude-Sonnet figures on the live site as historical data rather than current reality.

Key Takeaways

  • Aider's pricing guide overstates costs by roughly a third by using Claude 3.7 rates instead of Claude Sonnet 5.5's $2/$10 pricing.
  • Anthropic launched Claude Sonnet 5.5 on September 28, 2026, with a 1M-token context window.
  • The guide's comparison to Cursor Pro is skewed against Aider due to the outdated model pricing.
  • BYOK tools require active monitoring of vendor price changes, as they lack the internal control of flat-rate services.

The Bottom Line

Documentation that ignores vendor rate cards is documentation that lies. If you're paying Aider's quoted rates, you're overpaying for the privilege of being misinformed.