Grok 4-Test-Time-Compute
Grok 4-Test-Time-Compute is an AI model published by xAI, released on 2025-07-10, for 推理大模型, with 0.0B parameters, and 132K tokens context length, under the 不开源 license.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Modality | Input | Output |
|---|---|---|
| Text | $3 | $15 |
| Image | $3 | -- |
Grok-4-Test-Time-Compute是Grok 4的深度推理模型，即使用更多的推理时间来获得更好的效果。
该模式的Grok-4在各方面的评测上都比基础版本有所提高。但是，会生成更多的推理tokens，也有人称之为Gro-4 (Reasoning)。
非推理版本也有人称为Grok（Standard）
欢迎关注 DataLearner 官方微信，获得最新 AI 技术推送
