Unverified50% confidenceFactExact time

GPT-4 使用 cl100k_base tokenizer，与 Claude 的 tokenizer 对中文、代码等内容的切分策略不同

1

Sources

50%

Confidence

Long-term

Relevance

8/22/2026

First Seen

Sources

Codex on AWS Bedrock Billing Bug: Why Bills Are Surging 10x and How to Respond

Related Entities

GPT-4model Claudemodel cl100k_baseconcept

Related Claims

UnverifiedGPT-4所使用的cl100k_base词表包含约10万个Token单元，对中文单字的覆盖率相对较低72% similar VerifiedThe mainstream GPT-4 has long remained in the 8K–32K token range.70% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/788286

API

curl https://kongchang.com/api/v1/knowledge/claims/788286

MCP

get_claim(id=788286)