模型详情
o3-pro
The o-series of models are trained with reinforcement learning to think before they answer and perform complex reasoning. The o3-pro model uses more compute to think harder and provide consistently better answers. Note that BYOK is required for this model. Set up here: https://openrouter.ai/settings/integrations
模型规格
- 上下文长度
- 200K
- 最大输出
- 100K
- 输入/输出模态
- 文本 / 图像
- 发布日期
- 2025-06
02
API 端点
单一 MixRoute 网关，OpenAI 兼容
-
OpenAI-compatible
/v1/responsesPOST
03
价格
固定费率，单位：/1M Tokens
输入
$20.0000 /1M Tokens
补全
$80.0000 /1M Tokens
05
选型摘要
快速判断是否适合你的工作负载
What is this model good for?
Complex multi-step reasoning with deep thinking
Use o3-pro for the hardest reasoning tasks across math, science, coding, and analysis. It allocates extra compute to think longer and produce more reliable answers. Note that BYOK is required.
What should you check before using it?
Set up BYOK and plan for higher latency
This model requires Bring Your Own Key (BYOK). Expect longer response times due to extended reasoning. Test representative workloads to confirm latency and cost fit your use case.
Why use it through MixRoute?
Use the compatible endpoint with stable model ID
Use the confirmed compatible endpoint with model ID o3-pro through MixRoute, then run the same integration tests used for the current client.
08
Token 成本估算
基于本页价格的实时估算，非实际账单
此模型没有缓存读取价格，不计算缓存费用
单一端点，可验证的决策
使用相同的请求格式测试此模型和备用路由
从真实工作负载开始，再由质量、总成本和故障条件决定是否接入生产流量。