xAI · AI model
Grok 4.20 Multi-Agent (Beta)
Beta xAI model in which multiple agents collaborate in parallel on deep research tasks; the reasoning effort setting selects 4 or 16 agents. 1M-token context, text and image input; $1.25/$2.50 per 1M input/output tokens; Batch API 20% off.
Last verified Preview 7 sources
Key specs
- Provider
- xAI
- Family
- Grok 4
- API identifier
- grok-4.20-multi-agent-0309
- Availability
- Preview
- Release date
- 2026-03-10
- License
- Proprietary
- Context window
- 1,000,000 tokens
- API access
- Yes
- Consumer app
- Unknown
Capabilities
Inputs
- Text input Yes
- Image input Yes JPG/JPEG or PNG, max 20 MiB per image, no limit on image count (models page)
- Audio input No Model page lists modalities as text and image input, text output
- Video input No Model page lists modalities as text and image input, text output
- File input Unknown
Outputs
- Text output Yes
- Image output No Text-only output per model page
- Audio output No Text-only output per model page
Features
- Reasoning Native Model page lists reasoning as supported; the reasoning effort parameter controls the number of agents (4 or 16), not reasoning depth
- Tool calling Partial Model page lists function calling, but the multi-agent docs state client-side/custom tools are not supported; built-in server-side tools and remote MCP tools are supported
- Web search Yes web_search and x_search built-in tools shown in multi-agent docs
"Unknown" means we have not verified this capability yet; it does not mean the model lacks it.
Pricing
Estimate your monthly cost →| Dimension | Tier | Price | Since | Verified | Source |
|---|---|---|---|---|---|
| Input tokens | base | $1.25 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing |
| ≥ 200K prompt tokens (≥ 200K input tokens) | $2.50 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing Long-context rate: once a request's prompt reaches 200K tokens, all tokens in the request are billed at this rate (pricing page). | |
| Cached input tokens | base | $0.20 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing |
| ≥ 200K prompt tokens (≥ 200K input tokens) | $0.40 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing Long-context rate: once a request's prompt reaches 200K tokens, all tokens in the request are billed at this rate (pricing page). | |
| Output tokens | base | $2.50 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing |
| ≥ 200K prompt tokens (≥ 200K input tokens) | $5.00 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing Long-context rate: once a request's prompt reaches 200K tokens, all tokens in the request are billed at this rate (pricing page). | |
| Batch input tokens | base | $1.00 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing Batch API: 20% off standard rates for this model (pricing page and model page); the discount also applies to cached and reasoning tokens. |
| ≥ 200K prompt tokens (≥ 200K input tokens) | $2.00 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing Batch API: 20% off standard rates for this model (pricing page and model page); the discount also applies to cached and reasoning tokens. Long-context rate. | |
| Batch output tokens | base | $2.00 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing Batch API: 20% off standard rates for this model (pricing page and model page); the discount also applies to cached and reasoning tokens. |
| ≥ 200K prompt tokens (≥ 200K input tokens) | $4.00 per 1M tokens | 2026-10-12 | 2026-10-25 | API Pricing Batch API: 20% off standard rates for this model (pricing page and model page); the discount also applies to cached and reasoning tokens. Long-context rate. |
Prices are shown in the currency the provider publishes; we never convert currencies.
Price history (10 entries)
| Dimension | Tier | Price | From | Until |
|---|---|---|---|---|
| Input tokens | base | $1.25 per 1M tokens | 2026-10-12 | current |
| Input tokens | ≥ 200K prompt tokens (≥ 200K input tokens) | $2.50 per 1M tokens | 2026-10-12 | current |
| Cached input tokens | base | $0.20 per 1M tokens | 2026-10-12 | current |
| Cached input tokens | ≥ 200K prompt tokens (≥ 200K input tokens) | $0.40 per 1M tokens | 2026-10-12 | current |
| Output tokens | base | $2.50 per 1M tokens | 2026-10-12 | current |
| Output tokens | ≥ 200K prompt tokens (≥ 200K input tokens) | $5.00 per 1M tokens | 2026-10-12 | current |
| Batch input tokens | base | $1.00 per 1M tokens | 2026-10-12 | current |
| Batch input tokens | ≥ 200K prompt tokens (≥ 200K input tokens) | $2.00 per 1M tokens | 2026-10-12 | current |
| Batch output tokens | base | $2.00 per 1M tokens | 2026-10-12 | current |
| Batch output tokens | ≥ 200K prompt tokens (≥ 200K input tokens) | $4.00 per 1M tokens | 2026-10-12 | current |
Benchmarks
| Benchmark | Setting | Score | Reported by | Measured | Verified | Source |
|---|---|---|---|---|---|---|
| Text Arena Arena score, higher is better | Overall; style control on; multi-agent | 1,471 | Third party | 2026-10-23 | 2026-10-25 | Text Arena leaderboard (Arena) |
Scores measure voter preference, which can be influenced by answer style and length. The leaderboard offers a style-control adjustment, so record which view was used. Scores shift with every update, so a value is only meaningful together with its leaderboard date. Entries marked "Preliminary" or with few votes have wide confidence intervals. The same model can appear several times with different reasoning-effort suffixes (for example -high or -max). Category views (Coding, Math, Creative Writing and others) are different rankings from Overall.
One row per result, without rank numbers: each result was measured under its own setting (effort level, harness, tools, test version), as stated by the source, and scores compare only within the same setting. Self-reported results come from the provider and have not been reproduced independently.
Rankings
- #21 of 42 in Best AI API as of 2026-10-26
Derived from our published ranking snapshots; each ranking explains its criteria and weights.
Compare Grok 4.20 Multi-Agent (Beta)
Guides
- AI token pricing explained: input, output, cached and batch rates
- Reasoning models explained: thinking modes, effort and cost
- Multimodal AI explained: image, audio, video and file inputs, and what they cost
- Context windows explained: limits, long-context pricing and recall
- How to choose an AI model: a practical decision framework
About Grok 4.20 Multi-Agent (Beta)
Grok 4.20 Multi-Agent went live on March 10, 2026 and is documented as a beta whose API interface may change. A leader agent synthesizes the agents' discussion and returns the final answer; by default only its tool calls and final response are returned. It supports built-in server-side tools such as web_search and x_search plus remote MCP tools, and works with the xAI SDK and Responses API only.
Strengths and weaknesses
Strengths
- Runs 4 or 16 collaborating agents per request, chosen via the reasoning effort parameter
- Built-in server-side tools (e.g. web search, X search, code execution, collections search) and remote MCP tools supported
- 1M-token context window
- Batch API supported at 20% off standard rates
Weaknesses
- Client-side function calling, custom tools and the max_tokens parameter are not supported
- Chat Completions API not supported
- Lowest default-tier (Tier 0) rate limits in xAI's text line-up (9 requests/s, 2.5M tokens/min); the 16-agent setup uses significantly more tokens
Sources
Last verified Page updated 2026-10-26