Claude Code's Efficiency Concerns Raise Questions
Researchers have discovered significant efficiency issues with Claude Code, a large language model developed by Meta AI. Claude Code consumes up to 85,000 tokens before a user even types a word, compared to 7,000 tokens for OpenCode. This excessive usage is attributed to cache inefficiency and additional costs from subagents. The findings have sparked concerns about the model's scalability and potential impact on users.
Key points
- Claude Code, a large language model developed by Meta AI, consumes up to 85,000 tokens before a user even types a word.
- This is significantly more than OpenCode, which uses around 7,000 tokens for the same task.
- The excessive usage is attributed to cache inefficiency and additional costs from subagents.
- Researchers found that Claude Code re-writes tens of thousands of prompt-cache tokens mid-session, leading to premium billing for cache writes.
- A production repository's instruction file adds an average of 20,000 tokens to every single request, further contributing to the high usage.
Meta AI's Claude Code has been at the center of a recent controversy surrounding its efficiency concerns. Researchers have discovered that the large language model consumes a staggering amount of tokens before a user even types a word. In a comparison with OpenCode, Claude Code used roughly 33,000 tokens of system prompt, tool schemas, and injected scaffolding before the prompt even arrived, while OpenCode used about 7,000.
The excessive usage is attributed to cache inefficiency and additional costs from subagents. Claude Code re-writes tens of thousands of prompt-cache tokens mid-session, leading to premium billing for cache writes. Furthermore, a production repository's instruction file adds an average of 20,000 tokens to every single request, further contributing to the high usage.
In addition, the use of subagents has been found to add significant costs. A small task that cost 121,000 tokens done directly cost 513,000 tokens when fanned out to two subagents. This has sparked concerns about the model's scalability and potential impact on users.
The findings have significant implications for the development and deployment of large language models. As researchers continue to explore the capabilities and limitations of these models, it is essential to address efficiency concerns to ensure that they are scalable and cost-effective.
Sources
The WireByte editorial team synthesises technology news from multiple primary sources, verifies the facts, and links every source. Articles are produced with AI assistance and reviewed under our editorial policy.