Grok 4.20
GrokFlagship
ThinkingTool UseVisionStructured Output
About this model
Grok 4.20 is xAI's newest flagship model with industry-leading speed and agentic tool calling capabilities. It combines the lowest hallucination rate on the market with strict prompt adherance, delivering consistently...
Performance Tier
Flagship
Grok 4.20 is a flagship model from Grok : the most capable in their lineup.
Best-in-class model from this provider. Highest performance across benchmarks, ideal for demanding tasks.
Pricing
This model is included in Elosia plansModerate
Moderate cost. A balanced choice for regular use without constant cap watching.
| Type | per 1M tokens |
|---|---|
| Input (prompt) | $2.00 |
| Output (completion) | $6.00 |
| Cache read | $0.200 |
Capabilities
Context Length2.0M
Max Output Tokens—
TokenizerGrok
Inputtext, image, file
Outputtext
Release DateMarch 31, 2026
Benchmarks
General Intelligence
MMLU
91.2%
GPQA Diamond
88.5%
Mathematics
MATH-500
Not reported
AIME 2025
93%
Programming
HumanEval
Not reported
SWE-bench Verified
81%
LiveCodeBench
79.4%
Reasoning
IFEval
83%
ARC-AGI-2
15.9%
Humanity's Last Exam
35%
Recommended Use Cases
CodingAnalysisResearchGeneral ChatCreative Writing
Strengths
- 4-agent internal architecture reduces hallucinations by 65% — #1 on AA-Omniscience (78%)
- Industry-leading 2M token context window for massive document analysis
- Native multimodal understanding (text, image, video) with real-time X data access
- #1 on IFBench (83%) and #2 on tau2-Bench (97%) for instruction following and agentic tool use
Limitations
- Verbose output (~54M tokens during evaluation vs ~13M average) — higher cost per query
- Weak abstract reasoning (ARC-AGI-2: 15.9%) compared to top competitors
- Smaller third-party ecosystem than OpenAI/Anthropic
Resources
This model may use your data for training
