The engineering world is obsessed with 'tokenmaxxing'β€”the belief that maximizing AI token usage directly correlates with productivity. But new data from Jellyfish, a developer productivity platform, suggests this metric is dangerously misleading. By analyzing 12,000 developers across 200 companies in Q1 2026, researchers found that while higher token counts do increase output, they do so at an exponentially higher cost per merged pull request.

The Cost of Power Usage

Token consumption varies wildly among developers. The median user consumes approximately 51 million tokens per month, costing about $52.38 based on Claude API pricing. However, the 90th percentile users burn through over 380 million tokens monthly, racking up a bill of roughly $691.14. This disparity highlights a small group of power users driving a disproportionate share of total expenditure, raising questions about the efficiency of such extreme usage patterns.

Output vs. Efficiency Trade-Off

When joined with actual developer output, the data reveals a stark reality. Developers in the bottom 20% of token spend used only about three dollars’ worth of tokens in a quarter yet shipped 11 merged PRs, resulting in a cost of just $0.28 per PR. In contrast, the top 20% spent $1,822 over the same period to ship 23 PRs, driving the cost per PR up to $89.32. While throughput increased, the return on investment plummeted.

The Agentic Barrier

The research indicates that tokens behave less like linear inputs and more like rocket fuel. To double your throughput, you might need to increase token usage tenfold. The median developer uses about 7 million tokens per PR, while top decile users consume 69 million. This inefficiency explains why many teams struggle to realize the promise of autonomous AI agents; the infrastructure costs for sandboxing and context engineering create an 'agentic barrier' that raw token spending cannot overcome.

Key Takeaways

  • Median token cost is ~$52/month, but top users spend ~$691/month for only double the PR output.
  • Cost per merged PR jumps from $0.28 for low-usage devs to $89.32 for high-usage devs.
  • Broad, moderate adoption delivers more organizational value than narrow, extreme usage.
  • Autonomous agents require infrastructure investment, not just higher token budgets.

The Bottom Line

Stop counting tokens and start measuring value. If you aren't optimizing for efficiency per PR, you're just burning cash on digital rocket fuel that barely moves the needle.