Unverified50% confidenceFactExact time
The same 1,000-character Chinese text produces approximately 800 tokens under GPT-4o's tiktoken (cl100k_base), approximately 750 under the Claude tokenizer, and approximately 900 under Llama's sentencepiece tokenizer.
1
Sources
50%
Confidence
Long-term
Relevance
8/17/2026
First Seen
Sources
Inferock Bench：为每次LLM调用开一张独立账单的开源成本审计工具
Related Claims
UnverifiedAmong mainstream models such as GPT and Claude, approximately every 750 English words or 500 Chinese characters corresponds to 1,000 tokens.78% similarUnverifiedClaude Code charges by token; in English approximately 4 characters correspond to 1 token, while in Chinese approximately 1–2 characters correspond to 1 token.77% similarUnverified128K tokens约对应10万字中文内容75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/764520API
curl https://kongchang.com/api/v1/knowledge/claims/764520MCP
get_claim(id=764520)