免费工具 · 无需注册
LLM Token 与 API 成本计算器
粘贴任意 prompt 即可数出它的 token 数,并立刻看到在 GPT-5.5、Claude、Gemini、GLM、DeepSeek、Qwen 等各模型上跑一次的精确成本(从最便宜排起)。这里每个模型都能用一个 apimodels.app API、一个 key 调用。
0 输入 token
模型回复长度的估计值。
Token 数用 GPT o200k 分词器;Claude / Gemini 略有差异(接近估算)。
| 模型 | 输入 $/1M | 输出 $/1M | 每次 | 合计 |
|---|---|---|---|---|
gpt-6-luna最便宜 OpenAI / Other · GPT-6 Luna | $0.09 | $0.45 | $0.00023 | $0.00023 |
claude-haiku-5-5 Claude · Claude Haiku 5.5 | $0.09 | $0.45 | $0.00023 | $0.00023 |
qwen3.8-flash OpenAI / Other · Qwen3.8 Flash | $0.15 | $0.47 | $0.00023 | $0.00023 |
gpt-5.6-luna OpenAI / Other · GPT-5.6 Luna | $0.18 | $0.96 | $0.00048 | $0.00048 |
deepseek-v4.1-flash OpenAI / Other · DeepSeek V4.1 Flash | $0.27 | $1.08 | $0.00054 | $0.00054 |
deepseek-v4-flash OpenAI / Other · DeepSeek V4 Flash | $0.27 | $1.08 | $0.00054 | $0.00054 |
gemini-3-flash-preview Gemini · Gemini 3 Flash Preview | $0.2 | $1.2 | $0.00060 | $0.00060 |
qwen3.7-plus OpenAI / Other · Qwen3.7 Plus | $0.36 | $1.44 | $0.00072 | $0.00072 |
claude-haiku-4-5-20251001 Claude · Claude Haiku 4.5 | $0.353 | $1.765 | $0.00088 | $0.00088 |
deepseek-v4-pro-0813 OpenAI / Other · DeepSeek V4 Pro 0813 | $0.993 | $1.985 | $0.00099 | $0.00099 |
gemini-3.8-flash Gemini · Gemini 3.8 Flash | $0.45 | $2.25 | $0.00113 | $0.00113 |
gemini-3.1-flash-lite Gemini · Gemini 3.1 Flash Lite | $0.375 | $2.25 | $0.00113 | $0.00113 |
gpt-6.1-terra OpenAI / Other · GPT-6.1 Terra | $0.5 | $2.5 | $0.00125 | $0.00125 |
gemini-3.6-flash Gemini · Gemini 3.6 Flash | $0.662 | $3.302 | $0.00165 | $0.00165 |
gpt-5.6-terra OpenAI / Other · GPT-5.6 Terra | $0.551 | $3.309 | $0.00165 | $0.00165 |
deepseek-v4-pro OpenAI / Other · DeepSeek V4 Pro | $1.247 | $3.742 | $0.00187 | $0.00187 |
glm-5.2 OpenAI / Other · GLM-5.2 | $1.26 | $3.96 | $0.00198 | $0.00198 |
gemini-3.5-flash Gemini · Gemini 3.5 Flash | $0.662 | $3.971 | $0.00199 | $0.00199 |
glm-5.3 OpenAI / Other · GLM-5.3 | $1.33 | $4.18 | $0.00209 | $0.00209 |
grok-4.5 OpenAI / Other · Grok 4.5 | $1.275 | $4.25 | $0.00213 | $0.00213 |
grok-4.6 OpenAI / Other · Grok 4.6 | $1.5 | $4.5 | $0.00225 | $0.00225 |
gpt-6.1-sol OpenAI / Other · GPT-6.1 Sol | $1 | $5 | $0.00250 | $0.00250 |
gpt-6-sol OpenAI / Other · GPT-6 Sol | $1 | $5 | $0.00250 | $0.00250 |
gemini-3.1-pro-preview Gemini · Gemini 3.1 Pro Preview | $0.882 | $5.294 | $0.00265 | $0.00265 |
claude-sonnet-4-5-20250929 Claude · Claude Sonnet 4.5 | $1.059 | $5.295 | $0.00265 | $0.00265 |
claude-sonnet-4-6 Claude · Claude Sonnet 4.6 | $1.059 | $5.295 | $0.00265 | $0.00265 |
gemini-4-argon Gemini · Gemini 4 Argon | $1.1 | $5.5 | $0.00275 | $0.00275 |
qwen3.8-max OpenAI / Other · Qwen3.8 Max | $2 | $6 | $0.00300 | $0.00300 |
claude-sonnet-5-5 Claude · Claude Sonnet 5.5 | $1.2 | $6 | $0.00300 | $0.00300 |
gpt-5.6-sol OpenAI / Other · GPT-5.6 Sol | $1.324 | $6.618 | $0.00331 | $0.00331 |
qwen3.7-max OpenAI / Other · Qwen3.7 Max | $2.25 | $6.75 | $0.00337 | $0.00337 |
claude-sonnet-5 Claude · Claude Sonnet 5 | $1.6 | $8 | $0.00400 | $0.00400 |
claude-opus-4-5-20251101 Claude · Claude Opus 4.5 | $1.765 | $8.824 | $0.00441 | $0.00441 |
claude-opus-4-6 Claude · Claude Opus 4.6 | $1.765 | $8.824 | $0.00441 | $0.00441 |
gemini-3-pro-preview Gemini · Gemini 3 Pro Preview | $1.6 | $9.6 | $0.00480 | $0.00480 |
gpt-5-4 OpenAI / Other · GPT-5.4 | $1.8 | $10.8 | $0.00540 | $0.00540 |
claude-opus-5-5 Claude · Claude Opus 5.5 | $2.4 | $12 | $0.00600 | $0.00600 |
claude-opus-4-8 Claude · Claude Opus 4.8 | $3 | $13 | $0.00650 | $0.00650 |
claude-opus-5 Claude · Claude Opus 5 | $3 | $15 | $0.00750 | $0.00750 |
gpt-6-astra OpenAI / Other · GPT-6 Astra | $3.4 | $17 | $0.00850 | $0.00850 |
gpt-5-5 OpenAI / Other · GPT-5.5 | $3 | $18 | $0.00900 | $0.00900 |
claude-opus-4-7 Claude · Claude Opus 4.7 | $3.676 | $18.382 | $0.00919 | $0.00919 |
claude-fable-5 Claude · Claude Fable 5 | $5 | $25 | $0.0125 | $0.0125 |
claude-fable-5-1 Claude · Claude Fable 5.1 | $5 | $25 | $0.0125 | $0.0125 |
价格为 apimodels.app 费率(每百万 token 美元),可能调整;thinking 版与基础模型同价。
用一个 key 调用以上任意模型——去创建 API key,几分钟即可开始。
获取 API key常见问题
token 是怎么计算的?
我们在浏览器里用 GPT o200k_base 分词器(GPT-4o / GPT-5 同款)对文本分词。Claude 和 Gemini 的分词略有不同,所以这两类的数字按接近的估算看待即可,通常相差 ~10-20%。
API 成本怎么算的?
成本 = 输入 token × 输入单价 + 输出 token × 输出单价。单价是 apimodels.app 的每百万 token 美元价(来自我们的实时模型表),只对成功请求扣费。
能对比哪些模型?
GPT-5.5 / GPT-5.4、Claude(Opus/Sonnet/Haiku)、Gemini(Pro/Flash),以及 GLM、DeepSeek、Qwen、MiniMax、Grok——全部用一个 API、一个 key 即可调用。