Grok 4.20 Experimental Beta Reasoning
At a glance
- Price per 1M tokens (input)
- $1.25
- Price per 1M tokens (output)
- $2.50
- Context
- 1,000,000 tokens
- Free access
- yes (see below)
Refreshed daily; data verified September 25, 2026. Published September 25, 2026.
Grok 4.20 Experimental Beta Reasoning by xAI is designed for developers and technical teams building applications that require multimodal understanding and autonomous reasoning. It excels at tasks involving image analysis, function execution through tool calling, and complex logical inference where step-by-step reasoning improves accuracy. The model distinguishes xAI's approach by emphasizing transparent reasoning traces and tight integration between vision and action-oriented capabilities, enabling more reliable automation in dynamic environments.
Specifications & pricing
| Input (per 1M tokens) | $1.25 |
|---|---|
| Output (per 1M tokens) | $2.50 |
| Cache read (per 1M tokens) | $0.20 |
| Context window | 1,000,000 tokens |
| Max output | 1,000,000 tokens |
| Capabilities | images, tool calling, reasoning |
LiteLLM community dataset (MIT), verified September 25, 2026. Official xAI pricing.
Where to try Grok 4.20 Experimental Beta Reasoning for free
- xAI offers a free chat — Grok (limited free access). A vendor's free chat may run a different model from the same family — the exact model is not guaranteed.
Frequently asked questions
What types of tasks is this model best suited for?+
This model is ideal for workflows that combine visual input with decision-making, such as interpreting diagrams to trigger specific actions, analyzing product images for inventory updates, or processing scanned forms to extract and validate data. Its strength lies in chaining perception with executable steps via tool calling.
How does tool calling work in this model, and why is it useful?+
Tool calling allows the model to invoke predefined functions — like querying a database or sending an API request — based on its understanding of input. This enables the model to go beyond text generation and perform real-world actions, making it suitable for building agents that operate in structured environments.
How does this model differ from other versions in the Grok series?+
Unlike earlier Grok variants focused primarily on language, this experimental release prioritizes reasoning depth and multimodal coordination. It allocates more internal processing to validating conclusions before acting, especially when vision and tool use are combined, which reduces errors in complex tasks.
What should users know about its current limitations?+
As an experimental beta, the model may exhibit inconsistent performance on highly abstract reasoning or novel tool combinations. Users should implement fallback mechanisms and validate outputs in critical applications, as the reasoning process, while more transparent, is still under active refinement.
Compare with others
Org chart: how to move your company onto AI
A practical map: which company roles and processes AI agents can take over, where to start, and in what order to roll it out.