Grok 4.20 is xAI's flagship model that offers industry-leading inference speed and reliable agentic tool calling for complex tasks. It combines extremely low hallucination rates with strict prompt adherence, delivering consistently truthful and precise responses.
Reasoning
View model card in Model Garden
Model ID
grok-4.20-reasoning
Launch stage
Preview
Supported inputs & outputs
- Inputs:
Text , Image - Outputs:
Text
Capabilities
- Supported
- Function calling Preview feature
- Structured output Preview feature
- Reasoning Preview feature
- Not supported
- Batch predictions Preview feature
Usage types
- Supported
- Fixed quota Preview feature
- Not supported
- Dynamic shared quota Preview feature
- Provisioned Throughput Preview feature
Versions
-
grok-4.20-reasoning - Launch stage: Preview
- Release date: April 14, 2026
Supported regions
Model availability
- Global
-
global endpoint
Quota limits
global endpoint:
- QPM: 100
- Input TPM: 540,000
- Output TPM: 80,000
- Context length: 200,000
Pricing
Non-Reasoning
View model card in Model Garden
Model ID
grok-4.20-non-reasoning
Launch stage
Preview
Supported inputs & outputs
- Inputs:
Text , Image - Outputs:
Text
Capabilities
- Supported
- Function calling Preview feature
- Structured output Preview feature
- Not supported
- Batch predictions Preview feature
- Reasoning Preview feature
Usage types
- Supported
- Fixed quota Preview feature
- Not supported
- Dynamic shared quota Preview feature
- Provisioned Throughput Preview feature
Versions
-
grok-4.20-non-reasoning - Launch stage: Preview
- Release date: April 7, 2026
Supported regions
Model availability
- Global
-
global endpoint
Quota limits
global endpoint:
- QPM: 100
- Input TPM: 540,000
- Output TPM: 80,000
- Context length: 200,000
Pricing
