Grok 4.20

Overview

Freshness note: Model capabilities, limits, and pricing can change quickly. This profile is a point-in-time snapshot last verified on February 22, 2026.

Grok 4.20 is xAI’s latest release in the Grok family, building on the Grok 4 foundation with targeted improvements to reasoning depth, coding accuracy, and agentic tool use. xAI continues to differentiate primarily through real-time data access via the X (formerly Twitter) platform and an emphasis on fewer alignment constraints in conversational settings.

The model lands in a competitive frontier landscape alongside Claude Opus 4.6, GPT-5, and Gemini 3.1. Where Grok carves out its niche is in workflows that benefit from live data integration and in xAI’s ecosystem of developer tools.

Capabilities

Grok 4.20 targets reasoning-heavy and tool-intensive workloads:

• Improved multi-step reasoning with better performance on complex planning, math, and logic tasks compared to Grok 4.
• Strong coding capabilities for generation, debugging, refactoring, and code review across mainstream languages.
• Enhanced agentic tool use with more reliable function calling and better recovery from tool errors in multi-turn loops.
• Real-time data integration — the model can access current information through X platform data and web search, reducing the staleness problem common to static training cutoffs.
• Large output capacity (131K tokens) for long-form generation and detailed analysis.

The real-time data angle remains Grok’s most distinctive feature relative to competitors, though its value depends heavily on the specific use case.

Technical Details

xAI positions Grok 4.20 as a general-purpose frontier model with a bias toward developer and agentic applications:

• 256K token context window, placing it between the 200K of Claude models and the 2M of Gemini.
• 131,072 max output tokens, the highest among major frontier models.
• Text input and output with function calling and tool-use support.
• Live data access through X platform integration and web search capabilities.

The model is available through the xAI API with standard chat completions and function-calling interfaces. It also powers conversational features within the X platform.

Pricing & Access

xAI’s pricing structure for Grok 4.20 varies by access tier and is not published as a simple per-token rate at the time of writing. Developers should consult the xAI API documentation for current pricing details.

Access options:

• xAI API (direct, with standard and enterprise tiers)
• X platform (integrated into premium subscriptions)
• Select third-party integrations

The xAI API follows OpenAI-compatible conventions, which simplifies migration for teams already using that interface pattern.

Best Use Cases

Grok 4.20 fits well in scenarios that align with its strengths:

• Research and analysis tasks that benefit from real-time data and current event awareness.
• Coding workflows requiring large context and long output for complex refactors or full-file generation.
• Agentic tool-use patterns where the model drives multi-step workflows with function calls.
• Applications built within or adjacent to the X platform ecosystem.
• Exploratory and conversational use cases where users want fewer refusals on edge-case topics.

Less ideal for teams that need the broadest cloud deployment options (Bedrock, Vertex) or multimodal input capabilities, where competitors currently have stronger offerings.

Comparisons

• Claude Opus 4.6 (Anthropic): Opus has a more mature enterprise deployment story (Bedrock, Vertex, direct API) and stronger multimodal support; Grok counters with real-time data access and a larger max output window.
• GPT-5 (OpenAI): GPT-5 has the broadest ecosystem and strongest developer tooling; Grok competes on reasoning benchmarks and differentiates through live data integration.
• Gemini 3.1 (Google): Gemini leads on context length (2M vs. 256K) and native multimodal input; Grok’s advantages are real-time data and the largest max output capacity among frontier models.