Grok 4.20 Multi-Agent
Grok 4.20 Multi-Agent is a large language model made by SpaceXSI, added to the catalog on 31 Mar 2026. It has a context window of 2,000,000 tokens and costs $1.25 per 1M input tokens and $2.50 per 1M output tokens. It accepts text, images and files and produces text. Typical uses: deep research, synthesising information from supplied materials and coordinated tool-use workflows.
Price checked ยท source: OpenRouter
What Grok 4.20 Multi-Agent is
SpaceXSI: Grok 4.20 Multi-Agent is a variant of Grok 4.20 built for collaborative, agent-based workflows. Its catalog description says multiple agents operate in parallel to conduct deep research, coordinate tool use, and synthesize information. The model accepts text, images, and files, and produces text. It is listed with reasoning support, as well as structured outputs and response-format support. These details describe a model suited to work in which information is gathered or supplied in different forms and then returned as written output. The facts do not specify particular tools or supported file formats.
It has a context length of 2,000,000 tokens and a maximum output of 1,800,000 tokens. The context limit gives it room to process extensive material together, while the output limit permits very long text responses. This scale can suit deep-research tasks that involve reviewing substantial input and combining information in a synthesized response. Parallel agent operation is a stated part of its design, making coordinated research and tool-use workflows a central fit. Text, image, and file inputs can be included in these workflows, although the catalog does not explain how each input type is interpreted. Structured-output and response-format support may be relevant when an application needs results arranged in a specified format; the catalog does not provide further detail on those options. The facts identify its intended workflow but do not name specific research domains or tools.
Key features
- Accepts text, images, and files; produces text
- Has a context length of 2,000,000 tokens
- Supports reasoning, structured outputs, and response formats
- Uses multiple agents in parallel for research, tool use, and synthesis
Typical uses
- Deep research
- Synthesising information from supplied materials
- Coordinated tool-use workflows
- Producing structured text responses
Grok 4.20 Multi-Agent pricing
US dollars, list price through OpenRouter.
| Item | Standard |
|---|---|
| Input, per 1M tokens | $1.25 |
| Output, per 1M tokens | $2.50 |
| Cached input read, per 1M tokens | $0.20 |
| Web search, per search | $0.005 |
| Input above 200,000 prompt tokens, per 1M | $2.50 |
| Output above 200,000 prompt tokens, per 1M | $5.00 |
No price change since it was added on 31 Mar 2026.
| What a task costs | Cost |
|---|---|
| Answer a short question500 in / 300 out | $0.0014 |
| Summarise a 50-page PDF25,000 in / 800 out | $0.033 |
| Review a 2,000-line source file30,000 in / 2,000 out | $0.043 |
| Read a 300-page contract set150,000 in / 1,500 out | $0.191 |
Grok 4.20 Multi-Agent specifications
| Model id | x-ai/grok-4.20-multi-agent |
|---|---|
| Version | grok-4.20-multi-agent-20260309 |
| Context window | 2,000,000 tokensAbout 1,500,000 English words |
| Max output | 1,800,000 tokens |
| Knowledge cutoff | 2025-09-01 |
| Added to catalog | 31 Mar 2026 |
| Input types | text, image, file |
|---|---|
| Output types | text |
| Tool calling | No |
| Structured outputs | Yes |
| Reasoning effort control | Yes |
| Content moderation | Not applied by provider |
Sample outputs
The same prompts are given to every model, and the unedited output is shown with the date it was produced. Upend does not publish leaderboard scores it did not run.
No output has been recorded for Grok 4.20 Multi-Agent yet.
Compare Grok 4.20 Multi-Agent
Side by side with other models: prices, context window, what each accepts and supports, and the same four tasks priced on both.