Reasoning

Grok 3 Mini is a lightweight, smaller thinking model. Unlike traditional models that generate answers immediately, Grok 3 Mini thinks before responding. It’s ideal for reasoning-heavy tasks that don’t demand extensive domain knowledge, and shines in math-specific and quantitative use cases, such as solving challenging puzzles or math problems.

Reasoning is only supported by grok-3-mini-beta and grok-3-mini-fast-beta.
The Grok 3 models grok-3-beta and grok-3-fast-beta do not support reasoning.

Key Features

Think Before Responding: Thinks through problems step-by-step before delivering an answer.

Math & Quantitative Strength: Excels at numerical challenges and logic puzzles.

Reasoning Trace: The model's thoughts are available via the reasoning_content field in the response completion object (see example below).

You can access the model's raw thinking trace via the message.reasoning_content of the chat completion response.

Control how hard the model thinks

The reasoning_effort parameter controls how much time the model spends thinking before responding. It must be set to one of these values:

low: Minimal thinking time, using fewer tokens for quick responses.

high: Maximum thinking time, leveraging more tokens for complex problems.

Choosing the right level depends on your task: use low for simple queries that should complete quickly, and high for harder problems where response latency is less important.

Usage Example

Here’s a simple example using Grok 3 Mini to multiply 101 by 3. Notice that we can access both the reasoning content and final response.

Sample Output

Let me calculate 101 multiplied by 3:
101 * 3 = 303.
I can double-check that: 100 * 3 is 300, and 1 * 3 is 3, so 300 + 3 = 303. Yes, that's correct.

Final Response:
The result of 101 multiplied by 3 is 303.

Number of completion tokens (input):
14

Number of reasoning tokens (input):
310

When to Use Reasoning

Use grok-3-mini-beta or grok-3-mini-fast-beta: For tasks that can benefit from logical reasoning (such as meeting scheduling or math problems). Also great for tasks that don't require deep domain knowledge about a specific subject (eg basic customer support bot).

Use grok-3-beta or grok-3-fast-beta: For queries requiring deep domain expertise or world knowledge (eg healthcare, legal, finance).

Key Features

Control how hard the model thinks

Usage Example

Sample Output

When to Use Reasoning

Request

Responses

Reasoning

Key Features#

Control how hard the model thinks#

Usage Example#

Sample Output#

When to Use Reasoning#

Request

Responses

Key Features

Control how hard the model thinks

Usage Example

Sample Output

When to Use Reasoning