UTILITY · PROMPT ESTIMATOR
How many tokens will a task prompt cost?一个任务的 Prompt，要花多少 Token？
Paste a prompt to see recommended models, token usage, and the cost to finish the task.粘贴 Prompt，查看推荐模型、Token 消耗和完成费用。
Model prices · live模型价格 · 实时读取
Task prompt任务 Prompt
Classification runs on the local text only; the model list and prices come from AIHubMix.分类只分析本地文本；模型列表和价格来自 AIHubMix。
Model catalog模型目录
Estimate settings估算设置 ·
Current pick当前推荐
Routing fit路由适配—
Context上下文—
Tokenizer分词方式—Heuristic estimate when the catalog lists no tokenizer当前目录未提供 tokenizer 时使用估算口径
Input tokens输入 Token—Prompt input range per round单轮 Prompt 输入区间
Agent roundsAgent 对话轮数—Based on the expected rounds按当前执行轮数估算
Output tokens输出 Token—Expected total output预计总输出
Total tokens总 Token—Input + output输入 + 输出
Input / output price输入 / 输出价格—USD per 1M tokens美元 / 百万 Token
Estimated cost to finish完成任务预估—At the current rounds and prices按当前轮数与价格计算
ROUTING LEADERBOARD
Recommended models推荐模型
How tokens are estimated估算口径
Input tokens use a character-based heuristic range; multi-round tasks accumulate under a simplified model where each round carries the prompt plus all prior output. System messages, tool definitions, images, and files are not counted.输入 Token 使用字符启发式区间；多轮任务按“每轮携带 Prompt 与此前输出”的简化模型累计。system message、工具定义、图片和文件均未计入。
How cost is estimated费用口径
Cost uses the input/output prices and context tiers returned by the AIHubMix API; cache hits, tool calls, and retries are not included.费用使用 AIHubMix API 返回的 input/output 价格和上下文 tier，不含缓存命中、工具调用与重试。
See EP1 measured token breakdown →查看 EP1 实测 Token 明细 →