Skip to content
Model infoModelsSnapshot: 15 Oct 2026

xAI / Grok

Grok 4.20

Text generation and tool use through the Grok API.

Language models (LLMs)
availableClosed weightsText generationAgentsReasoning
Official documentation (opens in a new tab)Reviewed 12 Oct 2026
14 of 20 specification fields collected
Context window1M1,000,000 tokens
Maximum outputNot collectedAwaiting source review
Input / 1M tokens$1.25xAI API
Output / 1M tokens$2.50USD · observed 12 Oct 2026

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
1,000,000 tokensSource for Context / input limit (opens in a new tab)
Maximum output
Not collected
Input modalities
Text, ImageSource for Input modalities (opens in a new tab)
Output modalities
TextSource for Output modalities (opens in a new tab)
Reasoning
SupportedSource for Reasoning (opens in a new tab)
Tool calling
SupportedSource for Tool calling (opens in a new tab)
Structured output
SupportedSource for Structured output (opens in a new tab)
Streaming
SupportedSource for Streaming (opens in a new tab)

SSE streaming is documented for text-output models.

Batch processing
SupportedSource for Batch processing (opens in a new tab)

Supported

Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
2026-03Source for Release date (opens in a new tab)

API availability announcement month from the provider changelog; exact day not collected.

Parameters
Not collected
Architecture
Not collected
Languages
Not collected
API aliases
View all 15 entries
  • grok-4.20-reasoning-latest
  • grok-4.20
  • grok-4.20-reasoning
  • grok-4.20-0309
  • grok-4.20-beta-0309-reasoning
  • grok-4.20-beta
  • grok-4.20-beta-0309
  • grok-4.20-beta-latest
  • grok-4.20-beta-latest-reasoning
  • grok-4.20-beta-reasoning
  • grok-4.20-experimental-beta-0304-reasoning
  • grok-4.20-experimental-beta-0304
  • grok-4.20-experimental-beta-reasoning-latest
  • grok-4.20-experimental-beta-latest
  • grok-4.20-reasoning-gv2
Source for API aliases (opens in a new tab)

Routing aliases may point to a newer release. They are not additional independent models.

Published availability regions
us-east-1, us-west-2Source for Published availability regions (opens in a new tab)

Published model regions are not a processing-location guarantee for the global endpoint.

Published request quota
37 requests/secondSource for Published request quota (opens in a new tab)

Published card quota; team-specific rate limits may differ.

Published token quota
10,000,000 tokens/minuteSource for Published token quota (opens in a new tab)

Published card quota; team-specific rate limits may differ.

Context caching
SupportedSource for Context caching (opens in a new tab)

Documented on at least one recorded endpoint. See Caching for host, lifecycle and billing restrictions.

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

xAI API

Provider source (opens in a new tab)
API model ID
grok-4.20-0309-reasoning
Access
API account required. Model access and quotas depend on the team and location.
Regions
Global endpoint; processing location is not guaranteed.
INPUT / 1M TOKENS$1.25
OUTPUT / 1M TOKENS$2.50

Global standard token rates. At 200,000 prompt tokens or more, higher rates apply to all tokens in the request. Tool charges are additional. Cached input price: $0.20 / 1M tokens (base tier).

Pricing by prompt length, USD / 1M tokens
Input lengthInputOutputCached input
Below 200,000 prompt tokens$1.25$2.50$0.20
200,000 or more prompt tokens$2.50$5.00$0.40
USD · observed 12 Oct 2026 · Pricing evidence (opens in a new tab)

Data & enterprise terms

Provider documents no training on API inputs or outputs without explicit permission, and default 30-day audit retention. Team-level zero data retention changes available features; consult the linked policy for scope and exceptions.

Read the service policy (opens in a new tab)

Perplexity Agent API · default

Provider source (opens in a new tab)
API model ID
xai/grok-4.20-reasoning
Access
Perplexity API key; POST https://api.perplexity.ai/v1/agent (compatibility alias: /v1/responses). Set service_tier=default. auto or an omitted tier also selects default processing.
Regions
Not collected
Streaming
SupportedSource for Streaming (opens in a new tab)
Context caching
SupportedSource for Context caching (opens in a new tab)
Service tier
defaultSource for Service tier (opens in a new tab)
INPUT / 1M TOKENS$1.25
OUTPUT / 1M TOKENS$2.50

Perplexity Agent API default processing; USD per million tokens. Long-context rates apply by prompt length; the higher tier starts at exactly 200,000 tokens. Tool calls are charged separately when used. Rates are host-specific, not the model developer’s direct-API prices. Unknown cache-write charges are not treated as zero. Cached input price: $0.20 / 1M tokens (base tier).

Pricing by prompt length, USD / 1M tokens
Input lengthInputOutputCached input
Below 200,000 prompt tokens$1.25$2.50$0.20
At least 200,000 prompt tokens$2.50$5.00$0.20
USD · observed 12 Oct 2026 · Pricing evidence (opens in a new tab)
Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
Tool add-onWeb search1$0.0025invocation
Tool add-onFast search2$0.001invocation
Tool add-onImage search3$0.0025successful invocation
Tool add-onURL fetch4$0.0005invocation
Tool add-onPeople search5$0.005invocation
Tool add-onFinance search6$0.005invocation
Tool add-onCode sandbox7$0.03container session
Tool add-onSearch inside sandbox8$0.0025SDK search request

Rate conditions

  1. Standard web search. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
  2. web_search with search_type=fast. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
  3. Image search bills successful invocations. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
  4. Fetch and extract URL content. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
  5. People search tool. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
  6. Financial search tool. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
  7. Up to 20 minutes of active use per billing window; this is not a runtime cap. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
  8. Additional SDK search calls made inside a sandbox. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

xAI API

grok-4.20-0309-reasoning
Documented
Support evidence (opens in a new tab)
How it works
Automatic prefix cachingSource for How it works (opens in a new tab)
Minimum reusable input
Not collected
Lifetime & refresh
No guaranteed duration. Entries can be evicted because of load or server restarts.Source for Lifetime & refresh (opens in a new tab)
How to enable it
Keep the initial messages unchanged. Use x-grok-conv-id, or prompt_cache_key in Responses, to improve routing consistency.Source for How to enable it (opens in a new tab)
Usage fields
Chat: usage.prompt_tokens_details.cached_tokens. Responses: usage.input_tokens_details.cached_tokens.Source for Usage fields (opens in a new tab)
Billing rules
Cached input uses the rate for the selected model and endpoint. Uncached input and output are billed separately; observe any prompt-length tiers.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Hits are best effort. Editing or reordering early messages breaks prefix reuse. Streaming and tool-call history are supported.Source for Limits & data boundaries (opens in a new tab)
Cache reads, writes & storage
ChargeApplies toRate / unit
Cache read1Text tokens · Below 200,000 prompt tokens$0.20 / 1M tokensPricing evidence (opens in a new tab)
Cache read1Text tokens · 200,000 or more prompt tokens$0.40 / 1M tokensPricing evidence (opens in a new tab)

Rate conditions

  1. Global standard token rates. At 200,000 prompt tokens or more, higher rates apply to all tokens in the request. Tool charges are additional.

Observed 12 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.

Caching reviewed 12 Oct 2026

Perplexity Agent API · default

xai/grok-4.20-reasoning
Documented
Support evidence (opens in a new tab)
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected

Do not inherit the model developer’s direct-API retention policy onto the Perplexity endpoint.

How to enable it
Not collected
Usage fields
Not collected
Billing rules
Published discounted cache-read rates apply to this default service tier. Write charges, minimum cacheable input, retention and cache controls have not been verified for this host.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Not collected
Cache reads, writes & storage
ChargeApplies toRate / unit
Cache read1Text tokens · Below 200,000 prompt tokens$0.20 / 1M tokensPricing evidence (opens in a new tab)
Cache read1Text tokens · At least 200,000 prompt tokens$0.20 / 1M tokensPricing evidence (opens in a new tab)

Rate conditions

  1. Perplexity Agent API default processing; USD per million tokens. Long-context rates apply by prompt length; the higher tier starts at exactly 200,000 tokens. Tool calls are charged separately when used. Rates are host-specific, not the model developer’s direct-API prices. Unknown cache-write charges are not treated as zero.

Observed 12 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.

Caching reviewed 12 Oct 2026

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

Developer-documented proprietary hosted Grok release. No public weights identified for this release in the reviewed documentation; public Grok 1 and Grok 2 base-weight releases are excluded.

Classification evidence (opens in a new tab)

Sources reviewed