Grok 4.20
grok-4-20grok-4.20- 🆕 xAI multimodal reasoning model, released March 2026.
- 📏 2-million-token context window for very long inputs.
- 🧠 Operates as a reasoning model that thinks before responding.
- 🔧 Agentic tool calling plus structured outputs supported.
- 👁️ Accepts text and image inputs, returns text.
- 🌐 Web search and function calling among native capabilities.
- ⚡ Tuned for fast responses in agentic, tool-calling workflows.
- 🏢 Developed by xAI.
xAI is an American artificial intelligence company and wholly owned subsidiary of SpaceX. The company develops AI systems under the Grok brand, spanning language models, image generation, and video synthesis. xAI has quickly established itself as a multimodal AI lab with…
Explore 8 more models by XAI (company) →Grok 4.20 is xAI's multimodal reasoning model, released in March 2026, built around agentic tool use and a 2-million-token context window. According to xAI's documentation, it operates as a reasoning model that "thinks before responding" for complex tasks, with web search and function calling among its native capabilities. The large context window is intended to support very long inputs and multi-step, tool-driven workflows rather than one-shot prompting.
The model accepts text and image inputs and returns text. Function calling and structured outputs are available through the API, letting developers connect Grok to external tools and request organized response formats. xAI's own docs describe Grok 4.20 as offering fast responses with agentic tool calling and strict prompt adherence.
Within the same family, xAI also offers [[sibling:grok-4-20-multi-agent|Grok 4.20 Multi-Agent]], a companion deployment released the same day. The lineage continued with the later [[sibling:grok-4-3|Grok 4.3]], released in April 2026, which extends the same generational line of reasoning models.
Grok 4.20 is offered through xAI's API on an on-demand, pay-per-token basis. As with other Grok models, image inputs are converted into tokens that count toward the shared context budget alongside text, keeping the full window available for combined multimodal prompts.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
| Seller | Reputation↓ | Routing | Input $/M | Cached $/M | Output $/M | Categories | API |
|---|---|---|---|---|---|---|---|
| Venice.ai Proxy 0x1f22…18c9 | 99 | #3 | $0.71 | $0.115 | $1.415 | chat,reasoning,vision,multimodal,web-search,x-search | openai-chat-completions |
| ▲ Apex Ant 0x73b4…e736 | 83 | #2 | $0.639 | $0.1346 | $1.2735 | chat,web-search,vision,multimodal,reasoning,long-context,agents | openai-chat-completions |
| Open Ant 0xe4f6…5bc4 | 74 | #4 | $0.923 | $0.1495 | $1.8395 | chat,reasoning,vision,multimodal,tools,long-context | openai-chat-completions |
| surplusintelligence.ai 0x0e49…8927 | 68 | #1 | $0.426 | $0.069 | $0.849 | agents,anon,chat,cheap,frontier,function-calling,multimodal,reasoning,research,tasks,tools,vision,web-search,x-search | openai-chat-completions |
| Super Seeder 0xd19f…41f3 | 40 | gated | $0.71 | $0.115 | $1.415 | chat,reasoning,vision,multimodal,tools,long-context | openai-chat-completions |
| Fire Ant 🔥🐜 0xbe05…bc5d | 37 | gated | $1.2671 | $0.207 | $2.5252 | chat,math | — |
| antseed-neon-puma-944e 0x6650…944e | 3 | gated | $0.4842 | $0.23 | $0.965 | chat,math | openai-chat-completions |
| antseed-opal-badger-2580 0xc85d…2580 | 1 | gated | $0.71 | $0.71 | $1.415 | chat | openai-chat-completions |
| Apex TEE Test 0xe672…7955 | 0 | gated | $1.1502 | $1.1502 | $2.2923 | chat,web-search,vision,multimodal,reasoning,long-context,agents | openai-chat-completions |
"Best price" and the seller table are live AntSeed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = on-chain trust (0-100). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on AntSeed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.