xAI: Grok 4.20 (Reasoning)

Overview

Grok 4.20 reasoning variant with a 1M-token context, focused on speed and agentic tool calling.

Provider

XAxAI

Modality

Text and image

Input

text, image

Output

text

Weights

Closed

Supported parameters

tools, structured_outputs, reasoningProvider's model page, checked October 8, 2026.

Distillable

NoLicence or terms, checked October 8, 2026.

Tags

reasoning, agents, vision, long-context

Docs

Provider documentation

Data policy

Prompts go to

xAI (SpaceXAI)

Retention

API requests and responses stored encrypted for 30 days for abuse auditing, then deleted, never used for training without explicit permission

Zero data retention

Available

Region

No region guarantee on the default endpoint, US regional endpoint (grok-4.7 and grok-4.6 only) at 1.1x pricing

Policy

xAI data policy

Summarised from the provider's published terms. The provider's own policy is what applies.

Where you can run it

How Grok 4.20 (Reasoning) can be deployed, bought and reached through Safeguard.

Deployment

Public cloudPrivate cloud

Procurement

Not published yet

Inference regions

Not published yet

Closed weights stay with the provider, so this model runs in the provider's cloud, never on-prem or air-gapped.

Pricing

Notes

Prompts <200K tokens. At or above 200K: $2.50 in / $5 out. Cached input $0.20. Batch 20% off.

Source

xAI pricing , last verified October 8, 2026

No markup on model usage: you pay the provider's list price.

API

OpenAI-compatible. Your API key and access to this model are set up for your workspace when you choose Use with Safeguard.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.safeguard.sh/gpt/v1",
    api_key="SAFEGUARD_API_KEY",
)

response = client.chat.completions.create(
    model="xai/grok-4.20-0309-reasoning",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Base URL https://api.safeguard.sh/gpt/v1, model xai/grok-4.20-0309-reasoning.