xAI pricing, complete breakdown

Verified 2026-05-25, cross-checked against xAI pricing page, litellm, openrouter

xAI's current model lineup is led by Grok 4.20, priced at $2.00 per million input tokens and $6.00 per million output tokens for both reasoning and non-reasoning modes. For high-velocity applications, Grok 4.1 Fast offers a significant discount at $0.20 per million input tokens. The newly released Grok 4.3 provides a mid-tier balance at $1.25 per million input tokens with a 1-million token context window. This page helps you navigate the trade-offs between per-token API costs and flat-rate subscription tiers.

Grok 4.1 Fast is the most economical entry point at $0.20 per million input tokens.

How xAI's pricing universe works

Frontier AI labs like xAI operate on a dual-revenue model to balance high compute costs with market penetration. They offer per-token API pricing for developers who need to scale programmatically, while providing flat-rate subscriptions like SuperGrok for consumers and teams who need predictable monthly costs. This multi-path strategy allows xAI to capture high-margin enterprise usage through the API and cloud marketplaces while maintaining a steady, low-churn subscriber base for their web and mobile interfaces.

API (per-token, metered)

For: Developers, technical teams, startups building products on top of Grok
  • Pay only for tokens consumed
  • Full model lineup including reasoning and fast variants
  • Programmatic access via xAI Console
When to use: When integrating xAI into your own product or running variable batch workloads
Best for: Builders with metered or unpredictable usage

Consumer subscriptions (SuperGrok Lite, SuperGrok)

For: Individuals using xAI directly for writing, coding, research, and media creation
  • Fixed monthly fee starting at $10.00
  • Increased usage limits and longer conversation history
  • Access to AI image and video creation tools
  • Expert mode for AI agents
When to use: When using xAI as a daily-driver AI assistant rather than building on it
Best for: Solo professionals, knowledge workers, and creative hobbyists

Business/Team plans (Grok Business)

For: Teams needing shared workspaces, admin controls, and data privacy
  • Per-seat billing at $30.00/month
  • Centralized billing and user analytics
  • Excluded from training by default
  • Domain verification and seat management
When to use: When deploying xAI across a team that requires collaborative features and administrative oversight
Best for: Mid-size organizations adopting AI for internal productivity

Enterprise (Grok Enterprise)

For: Large organizations with procurement requirements and compliance needs
  • Custom pricing and unlimited users
  • Single sign-on (SSO) and SCIM directory sync
  • Custom data retention policies
  • Dedicated onboarding and support
When to use: When security, compliance, and custom organizational structures are mandatory
Best for: Enterprises with procurement-led adoption

Cloud marketplaces (AWS, Google, Azure)

For: Organizations with existing cloud commits or strict data-residency requirements
  • Same models, often with price parity
  • Counts toward existing cloud spend commits (EDP/MACC)
  • Stays within the cloud provider's security boundary
When to use: When you prefer a single bill from your cloud provider and need to burn down existing commits
Best for: Cloud-committed enterprises
Which one should you pick? If you are building a software product, use the API for metered scaling. For personal use or creative tasks like image generation, SuperGrok Lite or SuperGrok provides the best value. Teams should opt for Grok Business to ensure data is excluded from training, while large organizations should contact sales for Grok Enterprise or deploy via cloud marketplaces for compliance.

🎁 Current promos and time-sensitive deals

What's active right now. Auto-hides expired items.
xAI Data Sharing Program
$150 in monthly API credits for users who consent to share API request metadata and outputs for model training.
expires expires_note · source
GSA OneGov Agreement
$0.42 per organization for an 18-month term for U.S. federal agencies.
expires 2027-03-25 · source
xAI incentivizes model improvement through a credit-back system for data sharing, effectively providing a free tier for developers willing to contribute to the Grok training set.

📅 What changed in the last 30 days

Populated from aicost_price_changelog. Hides automatically when no recent events.
·
·
·
·
·
·
·
·
·
·
·
·
·
·
·
·
·
·
·
·
·
·

Every xAI product, profiled

For each product, what it's for, who picks it, what to watch out for, pros and cons, and what we tell our consulting clients.

developer api

xAI API

Per-token (see API rates above)
Usage-based pricing (Pay-as-you-go) with rates ranging from $0.20 to $15.00 per million tokens.
Target users
software engineers, ai researchers, enterprise developers
Typical uses
  • Integrating Grok-4 reasoning into custom applications
  • High-speed text generation using Grok-4-1-fast
  • Real-time data processing with 128K context windows
  • Multi-agent system orchestration
Why pick it
Provides direct access to the Grok-4 model family with industry-leading reasoning capabilities and low-latency 'fast' variants.
Key features
  • Access to Grok-4, Grok-4-20, and Grok-4-1-fast models
  • 128,000 token context window support
  • Native tool use and function calling
  • Reasoning and non-reasoning model variants
  • Server-sent events (SSE) for response streaming
  • Data Sharing Program ($150 credit incentive)
⚠ Marketing gimmicks to watch
Usage Guideline Violation Fee
xAI reportedly charges $0.05 for every request blocked by safety filters before generation starts.
Impact: Pre-screen prompts with local moderation tools to avoid recurring fees for rejected inputs.
Special Token Overhead
The API appends pre-defined system tokens to requests which are included in the billable count but not shown in basic tokenizers.
Impact: Budget for a 5-10% token overhead beyond your local tokenizer estimates.
Pros
  • Highly competitive pricing for 'fast' model variants ($0.20/$0.50)
  • Generous $150 monthly credit for opting into data sharing
  • Strong reasoning performance on complex logic tasks
Cons
  • Safety filter fees can penalize developers for user-generated prompt violations
  • Token counting can be opaque due to system-added tokens
  • Limited regional availability compared to hyperscalers
Insider view
The xAI API is currently a 'best-of-both-worlds' play. The Grok-4-1-fast model is priced aggressively to undercut GPT-4o-mini and Claude Haiku, while the reasoning models offer a legitimate alternative to OpenAI's o1 series. The $150 credit for data sharing is the most aggressive developer acquisition tactic in the current market.
Max bang for buck
Enable the 'Share API Inputs' setting to receive the $150 monthly credit, which effectively makes low-volume development free.
🔒 Training-on-your-data policy
By default, API data is not used for training. Users can opt-in via the 'Data Sharing Program' to receive credits in exchange for training rights. Refer to https://x.ai/privacy.
🔄 Migration path
Upgrade when:
Your application requires dedicated capacity or SOC 2 compliance (Enterprise).
Downgrade when:
Latency is less critical than cost; switch from Grok-4-20 to Grok-4-1-fast.
Switch vendor when:
You require native multimodal (video/audio) inputs not yet supported by Grok.
ScenarioMonthlyAnnualNotes
Small startup using 50M fast tokens/mo $35 $420 Assumes 40M input ($8) and 10M output ($5) plus buffer, minus $150 credit if opted-in (resulting in $0 cost for this volume).
Mid-sized app using 500M fast tokens/mo $350 $4,200 Calculated at $0.20/$0.50 rates with a 4:1 input/output ratio.
consumer

SuperGrok Lite

$10/mo · $99.96000000000001/yr
$10/mo monthly billing, $8.33/mo billed annually
Target users
individual users, professionals
Why pick it
Auto-derived from subscription_plans (DB row id=373). Full editorial narrative pending.
Key features
  • 2x longer conversations in Chat
  • 1x AI agent on Expert mode
  • Try out AI image & video creation
  • Increased limits at regular speed
Insider view
[Editorial content pending — auto-stub from DB row. Narrative will be added via screenshot transcript ingestion or hand-authoring.]
ScenarioMonthlyAnnualNotes
Standard annual cost $10 $99.96 Lower rate with annual commitment ($8.33/mo)
consumer

SuperGrok

$30/mo · $300/yr
$30/mo monthly billing, $25/mo billed annually
Target users
individual users, professionals
Why pick it
Auto-derived from subscription_plans (DB row id=374). Full editorial narrative pending.
Key features
  • 5x longer conversations in Chat
  • 4x AI agents on Expert mode (collaborating to get you the best answers)
  • More usage, at lightning-fast speed (with HD 720p, 30-second video)
  • Upload more files for smarter help
  • Lightning-fast replies
Insider view
[Editorial content pending — auto-stub from DB row. Narrative will be added via screenshot transcript ingestion or hand-authoring.]
ScenarioMonthlyAnnualNotes
Standard annual cost $30 $300 Lower rate with annual commitment ($25/mo)
team

Grok Business

$30/mo · $300/yr
$30/mo monthly billing, $25/mo billed annually (per user)
Target users
small teams, collaborative teams
Why pick it
Auto-derived from subscription_plans (DB row id=375). Full editorial narrative pending.
Key features
  • Everything in SuperGrok
  • Sharing and collaboration
  • Centralized billing and invoicing
  • Advanced team + seat management
  • User analytics and reporting
  • Domain verification
  • Excluded from training by default
Insider view
[Editorial content pending — auto-stub from DB row. Narrative will be added via screenshot transcript ingestion or hand-authoring.]
ScenarioMonthlyAnnualNotes
Standard annual cost $30 $300 Lower rate with annual commitment ($25/mo)
enterprise

Grok Enterprise

Per-token (see API rates above)
Custom enterprise pricing — contact sales (per user)
Target users
large organizations, enterprise buyers
Why pick it
Auto-derived from subscription_plans (DB row id=376). Full editorial narrative pending.
Key features
  • Unlimited users
  • Single sign-on (SSO)
  • Directory sync (SCIM)
  • Custom role-based access controls
  • Custom data retention
  • Flexible organizational structures
  • Dedicated onboarding and support
Insider view
[Editorial content pending — auto-stub from DB row. Narrative will be added via screenshot transcript ingestion or hand-authoring.]

All xAI products at a glance

Scroll up to the product profile for full detail

ProductPriceBest forHeadline featureYearly estimate
SuperGrok Lite $10/mo Casual personal use X Platform Integration $100 (Annual plan)
SuperGrok $30/mo Power users Flagship Model Access $300 (Annual plan)
Grok Business $30/mo/user Small teams Admin Console $300/user (Annual plan)
xAI API Usage-based App development Function Calling Variable ($500+ typical)
Grok Enterprise Custom Large corporations SSO & Compliance Contact Sales

xAI vs the field

Same-tier comparison across top 5 vendors

Comparison tierAnthropicOpenAIGooglexAIVerdict
Flagship API (per 1M tokens)
Claude Opus 4.7
$25.00 (Out)
GPT-5.4
$15.00 (Out)
Gemini 3.1 Pro
$12.00 (Out)
Grok-3 (xAI API)
$15.00 (Out)
xAI matches OpenAI's flagship output pricing, while remaining significantly cheaper than Anthropic's Opus tier.
Small/Flash API (per 1M tokens)
Claude Haiku 4
$0.25 (In)
GPT-5.4 Nano
$0.20 (In)
Gemini 3.1 Flash-Lite
$0.25 (In)
Grok-mini (xAI API)
$0.20 (In)
xAI and OpenAI are currently tied for the lowest entry price in the 'Nano/Mini' model category.
Consumer Subscription
Claude Pro
$20/mo
ChatGPT Plus
$20/mo
Gemini Advanced
$20/mo
SuperGrok
$30/mo
xAI's flagship consumer tier carries a 50% premium over the standard $20/mo market rate set by competitors.

🌳 Which xAI product fits you?

3 questions, 1 recommendation
How do you intend to use Grok?
Recommended
xai api
The API is the only path for developers to integrate Grok into their own applications with pay-as-you-go billing.
See full profile ↑
Recommended
grok business + grok enterprise
Business provides the necessary administrative tools and user management for professional team environments.
See full profile ↑
Recommended
grok supergrok
SuperGrok provides the highest rate limits and access to the most capable models for power users.
See full profile ↑
Recommended
grok supergrok lite
Lite is the most cost-effective way to access Grok's core capabilities without the flagship price tag.
See full profile ↑

xAI is currently moving toward a unified pricing model across its Grok 4 family, simplifying the previous multi-tier structure.

→ Grok 4.20 (reasoning) — $2 / $6 per MTok now
No notable changes detected for grok-4-20-reasoning in the recent changelog window.
→ Grok 4.20 (non-reasoning) — $2 / $6 per MTok now
No notable changes detected for grok-4-20 in the recent changelog window.
→ Grok 4.1 Fast (reasoning) — $0.2 / $0.5 per MTok now
No notable changes detected for grok-4-1-fast-reasoning in the recent changelog window.
→ Grok 4.1 Fast (non-reasoning) — $0.2 / $0.5 per MTok now
No notable changes detected for grok-4-1-fast in the recent changelog window.
↓ Grok 4 (legacy) — $3 / $15 per MTok now
Input costs dropped from $3.00 to $1.25, while output costs saw a massive reduction from $15.00 to $2.50.
→ Grok 4.3 — $1.25 / $2.5 per MTok now
No notable changes detected for grok-4-3 in the recent changelog window.
→ grok-4-3 — $1.25 / $2.5 per MTok now
No notable changes detected for grok-4-3 in the recent changelog window.
Overall: xAI has transitioned from a tiered pricing strategy to a unified 'flat-rate' model for its primary Grok 4 lineup. While this increased costs for 'Fast' model users, it drastically lowered the barrier for flagship and vision capabilities.

API or subscription: which is cheaper for you?

Cross-over math at current rates

grok-supergrok-lite ($10/mo) vs grok-4-1-fast API ($0.2/$0.5 per MTok)
Break-even: ~55,555 messages/month (avg ~600 tokens each)

At a cost of $0.00018 per average message (400 input/200 output), the Lite subscription only saves money if you exceed 55,000 messages monthly.

👉 API is significantly cheaper for casual users; Lite subscription is only for power-users of the X platform interface.
grok-supergrok ($30/mo) vs grok-4-20 API ($2/$6 per MTok)
Break-even: ~15,000 messages/month (avg ~600 tokens each)

The flagship Grok-4-20 model costs $0.002 per average message via API. You need to send 15,000 messages a month to justify the $30 subscription.

👉 API is the better value for almost all users unless they require the native X.com integration features.
grok-supergrok ($30/mo) vs grok-4 API ($3/$15 per MTok)
Break-even: ~7,142 messages/month (avg ~600 tokens each)

Using the legacy Grok-4 model at $0.0042 per message, the breakeven point drops to roughly 7,142 messages per month.

👉 Heavy researchers using legacy models may find the subscription predictable, but API remains more flexible for variable usage.
Rule of thumb
xAI's API pricing is aggressively low compared to subscription costs. Subscriptions are primarily a 'convenience tax' for using the X.com interface rather than a cost-saving measure for high-volume token consumption.

🧮 Estimate your annual xAI cost

Pick a profile, see the all-in annual estimate

All estimates use 2026-05-25 rates. API rates verified against LiteLLM.

Current pricing (all production models)

ModelInput $/MOutput $/MCached $/MContext
Grok 4.20 (reasoning)
grok-4-20-reasoning		 $2 $6 $0.50 2,000,000
Grok 4.20 (non-reasoning)
grok-4-20		 $2 $6 $0.50 2,000,000
Grok 4.1 Fast (reasoning)
grok-4-1-fast-reasoning		 $0.20 $0.50 $0.050 2,000,000
Grok 4.1 Fast (non-reasoning)
grok-4-1-fast		 $0.20 $0.50 $0.050 2,000,000
Grok 4 (legacy)
grok-4		 $3 $15 $0.75 2,000,000
Grok 4.3
grok-4-3		 $1.25 $2.5 — 1,000,000
grok-4-3
grok-4-3		 $1.25 $2.5 — 1,000,000

Prices are in USD per 1 million tokens. Prompt caching is available for most models at a 75% discount. Verified as of 2026-05-25.

Full rate breakdown (all variants)

Variants beyond standard API: batch (async, 50% off), cached read (0.1x), cache writes (1.25x or 2x base), long-context tier (~2x above threshold).

Grok 4.20 (reasoning) grok-4-20-reasoning

Deep chain-of-thought for complex logic and scientific discovery
Primary useAdvanced mathematical proofs, complex architectural planning, and multi-step reasoning tasks.
Who picks itResearch engineers and developers building high-reliability autonomous systems.
Vs other xAI modelsPriced at $2/M input and $6/M output, it matches the non-reasoning version's cost but prioritizes logical depth over speed.
When to useUse when accuracy in logic is non-negotiable; switch to Grok 4.1 Fast for high-volume, simpler tasks.
Equivalents at other vendors
mistral
Mistral Large 3 Matches the $2/$6 pricing structure for high-tier reasoning and general intelligence.
google
Gemini 3.1 Pro Similar $2 input rate for deep reasoning tasks, though Grok offers a more competitive $6 output rate.

Grok 4.20 (reasoning) grok-4-20-reasoning

VariantInput $/MOutput $/MNotes
Standard $2 $6 Default per-token API rate
Cached read $0.50 $6 Cached prompt input (~0.1x base); output rate unchanged

Grok 4.20 (non-reasoning) grok-4-20

High-performance general intelligence for massive context windows
Primary useLarge-scale document synthesis and complex instruction following without extended chain-of-thought.
Who picks itEnterprise developers processing massive internal knowledge bases.
Vs other xAI modelsAt $2/M input and $6/M output, it offers the same 2M context window as the reasoning variant but with faster time-to-first-token.
When to useBest for RAG over 2M tokens where reasoning overhead isn't required; use Grok 4.3 for better price-to-performance on smaller contexts.
Equivalents at other vendors
mistral
Mistral Large 3 Identical $2/$6 pricing for general-purpose high-tier tasks and large-scale processing.
openai
GPT-5.4 Competes in the flagship tier, though Grok is cheaper than GPT-5.4's $2.5/$15 rates.

Grok 4.20 (non-reasoning) grok-4-20

VariantInput $/MOutput $/MNotes
Standard $2 $6 Default per-token API rate
Cached read $0.50 $6 Cached prompt input (~0.1x base); output rate unchanged

Grok 4.1 Fast (reasoning) grok-4-1-fast-reasoning

Sub-second reasoning for interactive agentic workflows
Primary useReal-time decision making and quick logical verification in chat applications.
Who picks itDevelopers building responsive AI agents that require logical consistency.
Vs other xAI modelsSignificantly cheaper than Grok 4.20 at $0.2/M input and $0.5/M output while maintaining the 2M context window.
When to useUse for low-latency reasoning tasks; switch to Grok 4.20 for PhD-level scientific complexity.
Equivalents at other vendors
deepseek
DeepSeek V3.2 (reasoner) Comparable $0.28/$0.42 pricing for fast reasoning and efficient logic processing.
openai
GPT-5.4 Nano Matches the $0.2 input rate for lightweight intelligence, though Grok provides a larger context window.

Grok 4.1 Fast (reasoning) grok-4-1-fast-reasoning

VariantInput $/MOutput $/MNotes
Standard $0.20 $0.50 Default per-token API rate
Cached read $0.050 $0.50 Cached prompt input (~0.1x base); output rate unchanged

Grok 4.1 Fast (non-reasoning) grok-4-1-fast

Efficient high-volume processing with massive context support
Primary useHigh-throughput classification, extraction, and summarization across long documents.
Who picks itData engineers and product teams scaling AI features cost-effectively.
Vs other xAI modelsPriced at $0.2/M input and $0.5/M output, it is the most economical way to access the 2M token context.
When to useBest for bulk processing where cost-efficiency is paramount; use Grok 4.1 Fast (reasoning) if logic checks are needed.
Equivalents at other vendors
deepseek
DeepSeek V3.2 (chat) Similar high-efficiency pricing at $0.28/$0.42 for high-volume chat and extraction.
openai
GPT-5.4 Nano Matches the $0.2 input rate for high-volume tasks with minimal latency.

Grok 4.1 Fast (non-reasoning) grok-4-1-fast

VariantInput $/MOutput $/MNotes
Standard $0.20 $0.50 Default per-token API rate
Cached read $0.050 $0.50 Cached prompt input (~0.1x base); output rate unchanged

Grok 4 (legacy) grok-4

Stable legacy endpoint for established production pipelines
Primary useMaintaining existing integrations that rely on specific Grok 4 behavior.
Who picks itTeams with validated prompts on Grok 4 not yet ready to migrate.
Vs other xAI modelsMost expensive at $3/M input and $15/M output; Grok 4.20 offers better performance at a lower price point.
When to useOnly for legacy compatibility; migrate to Grok 4.20 or 4.3 for significantly better unit economics.
Equivalents at other vendors
anthropic
Claude Opus 4.5 Similar premium tier positioning, though Grok 4 is cheaper than the $5/$25 Opus rates.
cohere
Command R+ 04-2024 Matches the $3 input rate for legacy high-tier models with large context capabilities.

Grok 4 (legacy) grok-4

VariantInput $/MOutput $/MNotes
Standard $3 $15 Default per-token API rate
Cached read $0.75 $15 Cached prompt input (~0.1x base); output rate unchanged

Grok 4.3 grok-4-3

Optimized mid-tier intelligence for standard context applications
Primary useEveryday automation, content generation, and structured data extraction.
Who picks itTeams requiring a balance of speed, cost, and 1M token context.
Vs other xAI modelsAt $1.25/M input and $2.5/M output, it provides a cost-effective alternative to the $2/M Grok 4.20.
When to useIdeal for standard workloads; use Grok 4.1 Fast if context is large but complexity is low.
Equivalents at other vendors
openai
GPT-5 Matches the $1.25 input price point for general-purpose use and reliable instruction following.
google
Gemini 2.5 Pro Similar $1.25/$10 pricing for professional workflows, though Grok offers much cheaper output.

Grok 4.3 grok-4-3

VariantInput $/MOutput $/MNotes
Standard $1.25 $2.5 Default per-token API rate
Standard $1.25 $2.5 Default per-token API rate

Subscription plans (consumer + business)

PlanAudienceMonthlyAnnualPer seatWhat's included
SuperGrok Lite consumer $10 $8.33/mo billed annually
($99.96/yr total)		 — 2x longer conversations in Chat · 1x AI agent on Expert mode · Try out AI image & video creation · Increased limits at regular speed
Limits: limits source: transcript: '2x longer conversations'; specific quotas not numerically published
grok.com ↗
SuperGrok consumer $30 $25/mo billed annually
($300/yr total)		 — 5x longer conversations in Chat · 4x AI agents on Expert mode (collaborating to get you the best answers) · More usage, at lightning-fast speed (with HD 720p, 30-second video) · Upload more files for smarter help · Lightning-fast replies
Limits: limits source: transcript: '5x longer conversations, 4x AI agents'
grok.com ↗
Grok Business team $30 $25/mo billed annually
($300/yr total)		 $1 Everything in SuperGrok · Sharing and collaboration · Centralized billing and invoicing · Advanced team + seat management · User analytics and reporting · Domain verification · Excluded from training by default
Limits: limits source: transcript: 'Everything in SuperGrok' + team features
grok.com ↗
Grok Enterprise Custom enterprise Contact — $1 Unlimited users · Single sign-on (SSO) · Directory sync (SCIM) · Custom role-based access controls · Custom data retention · Flexible organizational structures · Dedicated onboarding and support
grok.com ↗

Subscription pricing is separate from per-token API rates above.

What changed in the last 30-90 days

How buyers think about xAI pricing

Each scenario below is interactive — tweak the inputs to see how the math changes for your workload.

Grok 4.3 as the cheapest flagship API in the top 5 Western vendors

vibe-coderdevelopersolopreneur

The problem: You need high-intelligence reasoning and large context windows but cannot justify the premium rates of other major flagship models. Scaling a production application on top-tier models often leads to unsustainable monthly bills.

What to do: Grok 4.3 provides a balanced performance profile at a significantly lower entry point than legacy flagship models.

Processing 1 million input tokens and 1 million output tokens on Grok 4.3 costs $3.75 total ($1.25 input + $2.50 output). For a high-volume app processing 100 million tokens monthly, this results in a $375 bill (as of 2026-05-25).

→ Grok 4.3 delivers flagship-grade intelligence for under $4 per million combined tokens (as of 2026-05-25).

Quick calc — adjust for your workload
Per request: —  ·  Monthly: —  ·  Annual: —
Open full calculator with caching, batch, charts →

SuperGrok 30 dollar monthly rate versus Claude Pro and ChatGPT Plus

vibe-codersolopreneur

The problem: You are deciding between premium chat subscriptions and need to know if the higher price point of xAI's consumer offering is justified for your workflow. Standard subscriptions usually hover around 20 dollars.

What to do: SuperGrok is best utilized when your workflow requires multi-agent Expert mode or deep integration with X platform data.

At $30 per month, SuperGrok represents a $10 to $13 premium over competitors like ChatGPT Plus ($20) or Claude Pro ($17). This investment is primarily for the bundled image and video generation tools and real-time X search capabilities (as of 2026-05-25).

→ Expect to pay a 50 percent premium over standard AI subscriptions to access xAI's multi-agent ecosystem (as of 2026-05-25).

Quick calc — adjust for your workload
Per request: —  ·  Monthly: —  ·  Annual: —
Open full calculator with caching, batch, charts →

The 0.05 dollar pre-generation violation fee when it adds up

developerenterprise

The problem: Applications handling unfiltered user-generated content (UGC) risk high costs from requests that trigger safety filters. Unlike other vendors who may refuse for free, xAI charges for these blocked attempts.

What to do: Implement local moderation or regex filters to catch policy violations before they reach the xAI API endpoint.

If an unmoderated app receives 10,000 requests that violate usage guidelines, xAI will bill $500 in violation fees ($0.05 per request). This is an additional cost beyond any successful token generation (as of 2026-05-25).

→ Safety filter triggers cost $50 per 1,000 blocked requests (as of 2026-05-25).

Quick calc — adjust for your workload
Per request: —  ·  Monthly: —  ·  Annual: —
Open full calculator with caching, batch, charts →

Grok Business at 30 dollars per seat versus alternatives

smbenterprise

The problem: Small teams need to choose a collaborative AI platform but face varying per-user costs. You need to justify the higher seat price of Grok Business compared to standard office suites.

What to do: Select Grok Business if your team relies on real-time social data or requires the Enterprise Vault for data isolation.

A team of 10 users costs $300 per month on Grok Business. This is $100 more per month than a 10-user Claude Team plan ($20/seat) and $160 more than a basic Google Workspace AI add-on ($14/seat) (as of 2026-05-25).

→ Grok Business carries a $30 monthly per-seat cost for team-wide access (as of 2026-05-25).

Quick calc — adjust for your workload
Per request: —  ·  Monthly: —  ·  Annual: —
Open full calculator with caching, batch, charts →

Collections RAG storage costs 4x file storage

developerenterprise

The problem: Storing large datasets for Retrieval-Augmented Generation (RAG) can lead to unexpected infrastructure bills if you use high-performance collection storage for inactive data.

What to do: Use Collections for active, frequently queried data and move cold data to standard File storage to reduce daily overhead.

Storing 10 GiB in a Collection costs $1.00 per day ($0.10/GiB/day), totaling $30 per month. Storing that same 10 GiB as standard Files costs $0.25 per day ($0.025/GiB/day), saving $22.50 monthly (as of 2026-05-25).

→ RAG Collections are 4 times more expensive to store than standard files (as of 2026-05-25).

Quick calc — adjust for your workload
Per request: —  ·  Monthly: —  ·  Annual: —
Open full calculator with caching, batch, charts →

Volume discounts & partner programs

Heads up — these are community-sourced and analyst-reported terms. Specific credit amounts, discount percentages, and program thresholds change frequently. Always verify current terms directly with xAI before relying on a specific number. Treat reported figures as ballpark, not contract language.

xAI Provisioned Throughput

Threshold: minimum 30-day commitment

Typical discount (reported): reportedly approximately $10 per day per unit

Benefits:

How to engage: Contact [email protected] or [email protected] with expected TPM and preferred models

Source: docs.x.aivendor_official · cited 2026-05-09

xAI Data Sharing Program

Threshold: consent to share API request metadata and outputs for model training

Typical discount (reported): $150 in monthly API credits

Benefits:

How to engage: Enable 'Share API Inputs for Model Training' in xAI Console Settings > Data Sharing

Source: aifreeapi.comcommunity · cited 2026-01-16

GSA OneGov Agreement (Grok for Government)

Threshold: U.S. federal agency, department, or bureau

Typical discount (reported): $0.42 per organization for an 18-month term

Benefits:

How to engage: Procure through GSA Multiple Award Schedule (MAS) channels

Source: gsa.govvendor_official · cited 2025-09-25

xAI Enterprise Tier

Threshold: varies by contract

Typical discount (reported): volume pricing available on request

Benefits:

How to engage: Contact [email protected] for a custom plan

Source: x.aivendor_official · cited 2026-05-25

Azure AI Foundry: Grok 4 Managed Service

Threshold: Azure subscription required

Typical discount (reported): reportedly $5.50 per million input / $27.50 per million output tokens

Benefits:

How to engage: Deploy via Azure AI Foundry model catalog

Source: azure.microsoft.comvendor_official · cited 2025-09-29

Google Vertex AI: Grok Model Garden

Threshold: Google Cloud Platform project with Vertex AI enabled

Typical discount (reported): usage-based pricing (PayGo)

Benefits:

How to engage: Access through Vertex Model Garden in the Google Cloud console

Source: cloud.google.comvendor_official · cited 2026-05-13

Multi-cloud availability

Cloud-marketplace terms change frequently. Model availability dates, pricing parity, and regional features can drift week to week. Verify with each cloud's pricing page (AWS Bedrock, Google Vertex, Azure AI Foundry) before architecting around specifics.
CloudModel availabilityPrice vs vendor-directReasons to pick
Microsoft Azure (Azure AI Foundry) Grok 4, Grok 4 Fast Reasoning, Grok 4 Fast Non-Reasoning, Grok 3, Grok 3 Mini Varies by commitment (Pay-as-you-go or Reserved PTUs)
  • Data remains within the Azure tenant for enterprise privacy and compliance
  • Unified billing and governance through the Microsoft Foundry platform
  • Seamless Provisioned Throughput Unit (PTU) portability across different models
  • Enterprise-grade security features including RBAC, private networking, and customer-managed keys

vertexaisearch.cloud.google.com ↗
Oracle Cloud Infrastructure (OCI) Grok 4.3, Grok 4.20, Grok 4.20 Multi-Agent, Grok 3, Grok 3 Mini Varies by contract (On-Demand and Dedicated options available)
  • Zero data retention endpoints offer an extra layer of protection for sensitive enterprise data
  • Direct integration with OCI's high-performance AI infrastructure used to train next-generation models
  • Enterprise-grade data governance and management capabilities
  • Optimized for large-scale inferencing and business process automation

vertexaisearch.cloud.google.com ↗
OpenCode (opencode.ai) Grok Code Fast 1, Grok Build, and other preferred Grok models No additional charge for existing SuperGrok or X Premium subscribers
  • Eliminates the need for separate API key management (XAI_API_KEY) via OAuth login
  • Zero entry barrier for developers already holding X platform subscriptions
  • Direct integration into a terminal-based coding agent for enhanced efficiency

vertexaisearch.cloud.google.com ↗
OpenRouter Grok 4.20 (Reasoning and Multi-Agent variants), Grok 4.1 Fast Reportedly follows standard API rates with access to beta variants
  • Access to specific dated or beta model variants (e.g., grok-4.20-0309-reasoning)
  • Standardized API interface for developers using multiple LLM providers simultaneously
  • Lower barrier to entry for high-volume agent workflows compared to enterprise cloud contracts

vertexaisearch.cloud.google.com ↗
GitHub Models Grok 3, Grok 3 Mini Available for free preview (limited time)
  • Easy experimentation for developers within the GitHub ecosystem
  • No infrastructure setup required for initial testing and evaluation

vertexaisearch.cloud.google.com ↗

Free credits & startup programs

Program details and credit amounts shift often. Apply directly through each program's official page for current values, eligibility windows, and application requirements.

xAI API Free Trial & Data Sharing Program

Reported value: $25 one-time signup credit plus $150/month recurring

Eligibility: New xAI Console accounts receive $25; recurring $150/month requires opting into the data sharing program and a minimum $5 lifetime spend

How to apply: Sign up at console.x.ai; enable data sharing in the Billing section to unlock recurring credits

Apply / learn more at aifreeapi.com ↗

Y Combinator AI Starter Pack (YC AI Stack)

Reported value: over $5,000 in credits for GPT/Claude/Grok

Eligibility: Students who attend a YC university event (starting Fall 2025)

How to apply: Redeem via email link sent after attending an eligible YC university event

Apply / learn more at ycombinator.com ↗

Microsoft for Startups Founders Hub

Reported value: up to $150,000 in Azure credits

Eligibility: Early-stage startups (typically less than 7 years old and under $10M in revenue)

How to apply: Apply through the Microsoft for Startups portal; credits can reportedly be applied to Grok models hosted on Azure AI Foundry

Apply / learn more at microsoft.com ↗

xAI Grok API Public Beta

Reported value: $25 of free API credits per month

Eligibility: Publicly available to all developers during the beta period

How to apply: Create an account at console.x.ai to receive monthly beta credits

Apply / learn more at x.ai ↗

Grok Build Early Beta

Reported value: included with SuperGrok Heavy subscription

Eligibility: Subscribers to the SuperGrok Heavy plan ($300/month)

How to apply: Download the Grok Build CLI and log in with a SuperGrok Heavy account

Apply / learn more at x.ai ↗

Vercel AI Gateway xAI Integration

Reported value: varies by Vercel plan (Pro/Enterprise)

Eligibility: Vercel Pro and Enterprise plan subscribers

How to apply: Access Grok Imagine and other xAI models via the Vercel AI Gateway using an xAI API key

Apply / learn more at vercel.com ↗

Pricing gotchas to watch

Most gotchas below were surfaced by community reports. Some may have been fixed, changed, or never been the user-facing issue they appeared. Verify against current vendor docs before architecting around a workaround.

Usage Guideline Violation Fee

xAI has reportedly introduced a $0.05 fee for every request that is blocked by their safety filters before generation begins. This fee applies to the Responses API and is intended to discourage prompts that violate usage policies.

Workaround: Pre-screen prompts with local moderation models or strict regex filters to ensure they comply with xAI's safety guidelines before sending them to the API.

Source: vertexaisearch.cloud.google.comblog_post · cited 2026-05-21

Token Counting Discrepancy via Special Tokens

The xAI tokenizer page and API may report a lower token count than what is actually billed. Inference endpoints automatically append 'pre-defined tokens' or 'special tokens' to help the system process requests, which are included in the final billable token count.

Workaround: Budget for a small percentage of overhead (approximately 5-10 tokens per message) beyond the tokenizer's estimate to account for system-added tokens.

Source: vertexaisearch.cloud.google.comvendor_docs · cited 2026-01-29

Prompt Cache TTL and Sparse-Traffic Eviction

While xAI performs automatic prompt caching, cache entries are not guaranteed and can be evicted at any time due to server load. Community reports indicate that cache TTL has reportedly been reduced from 1 hour to as little as 5 minutes, making caching less effective for low-traffic applications.

Workaround: Use the 'x-grok-conv-id' header (Chat API) or 'prompt_cache_key' (Responses API) to implement sticky routing, which increases the likelihood of hitting the same server where the cache is resident.

Source: vertexaisearch.cloud.google.comblog_post · cited 2026-05-10

Flat Fees for Tool Invocations

Beyond token costs, xAI charges flat fees for invoking specific tools. Web Search and Code Execution calls are reportedly billed at $5.00 per 1,000 calls, while File Attachments carry a higher fee of $10.00 per 1,000 calls.

Workaround: Batch tool-heavy requests or use local processing for simple code execution tasks to avoid the per-invocation flat fee.

Source: vertexaisearch.cloud.google.comblog_post · cited 2026-05-01

Regional Payment Restrictions for India

xAI currently cannot process Indian payment cards for its API service due to regulatory requirements. Users in India are reportedly restricted to purchasing prepaid credits via 'Guest Checkout' or must use third-party providers.

Workaround: Use a third-party API aggregator or a non-Indian payment method if available to ensure uninterrupted service.

Source: vertexaisearch.cloud.google.comvendor_docs · cited 2026-02-04

Data Sharing Credit Program Requirements

xAI offers up to $150 per month in free API credits, but this is contingent on enabling the 'Share API Inputs for Model Training' toggle in the console. This program is subject to change and may not be available in all regions.

Workaround: Regularly check the 'Data Sharing' settings in the xAI console to ensure the program is active and credits are being applied.

Source: vertexaisearch.cloud.google.comblog_post · cited 2026-01-16

Hidden costs (25-40% beyond per-token rates)

Typical overhead: 25-40% beyond raw per-token rates.

What it costs to leave xAI

Migrating away from xAI involves transitioning from their specific tool-calling syntax and managing the loss of unique X platform data integrations. While the API is largely compatible with standard REST patterns, the Enterprise Vault and specific RAG Collection formats may require data re-indexing when moving to another provider.

Who is this for?

For vibe coders & solo devs

For rapid prototyping, Grok 4.1 Fast is your most efficient tool, offering input at $0.20 per million tokens. You should leverage the $150 monthly credit by enabling data sharing if your project allows for public training data. This effectively makes early-stage development free for most small-scale agent experiments. Be mindful of the $0.05 violation fee when testing edgy prompts.

For SMBs and growing teams

Small businesses should evaluate the Grok Business tier primarily for its Enterprise Vault and data isolation features. If you are already paying for X Premium, check if the Grok Build or OpenCode integrations can offset your need for separate API keys. The $150 monthly credit program is a significant subsidy for internal tool development. Avoid using Collections for long-term archiving to keep storage costs at the $0.025 per GiB rate.

For enterprise buyers

Enterprise buyers should look toward Provisioned Throughput for predictable latency, which reportedly starts with a 30-day commitment. If you are already on Azure or OCI, deploying Grok through their model gardens provides unified billing and potentially better data governance. For federal agencies, the GSA OneGov agreement offers a unique entry point at $0.42 per organization. Ensure your SOC 2 requirements are met through the Enterprise tier's dedicated infrastructure.
Need help deciding which xAI tier or model fits your workload? Book a $19.99 quick call →

Sources verified for this page

Primary: xAI pricing page

View all 23 cited insider sources across 10 domains

Generator: gen-v5.0.8-2026-05-25 · Last refreshed: Mon May 25 2026 17:43:40 GMT-0400 (Eastern Daylight Time) · Pricing snapshot: Mon May 25 2026 00:00:00 GMT-0400 (Eastern Daylight Time)

📖 Data sources & methodology 161 text models · 9 embeddings · 24 vision · 41 audio · 8 vector DBs across 10 vendor pages · last verified 2026-06-05

Methodology

  • All prices are USD per 1 million tokens, current as of 2026-06-05.
  • Vendor-published values have no mark. Inferred/extrapolated values are marked with * and listed below.
  • Batch API discounts are 50% off standard rates across providers that offer Batch mode.
  • Prompt caching discounts vary by provider (typically 80-90% off cached input tokens).
  • Regional data-residency surcharges (Anthropic 1.1x, OpenAI 1.1x, Google regional tiers) are NOT included in base rates.
  • Long-context pricing tiers apply when input exceeds model threshold.
  • Embedding prices are input-only (no output tokens generated).

Primary sources

Last-verified date is the most recent successful daily snapshot (aicost_pricing_snapshots) or, when no snapshot exists yet, the latest successful crawler run (aicost_crawler_runs). 10 of 10 vendors are currently verified. Aggregator services (TokenCost, AI Pricing Guru, etc.) are not listed.

Anthropic
2026-06-05
https://www.anthropic.com/pricing
Daily snapshot since Sep 2023 · 578 days captured
Anthropic Docs
2026-06-05
https://platform.claude.com/docs/en/about-claude/pricing
Daily snapshot since Sep 2023 · 578 days captured
OpenAI
2026-06-05
https://openai.com/api/pricing/
Daily snapshot since Sep 2023 · 579 days captured
Google AI
2026-06-05
https://ai.google.dev/gemini-api/docs/pricing
Daily snapshot since Dec 2023 · 554 days captured
Google Vertex
2026-06-05
https://cloud.google.com/vertex-ai/generative-ai/pricing
Daily snapshot since Dec 2023 · 554 days captured
DeepSeek
2026-06-05
https://api-docs.deepseek.com/quick_start/pricing
Daily snapshot since May 2024 · 493 days captured
xAI
2026-06-05
https://x.ai/api
Daily snapshot since Nov 2024 · 411 days captured
Mistral
2026-06-05
https://mistral.ai/pricing
Daily snapshot since Dec 2023 · 552 days captured
Cohere
2026-06-05
https://cohere.com/pricing
Daily snapshot since Sep 2023 · 578 days captured
Voyage AI
2026-06-05
https://docs.voyageai.com/docs/pricing

Inferred values (marked with * in calculator tables)

Derived from industry conventions, not directly published by the vendor. Typical conventions: cached input = 10% of base (90% off), Batch API = 50% of base (50% off).

Vendor / Model Field Why it’s inferred
Anthropic — Claude Sonnet 4.6 cachedInput Derived at 10% of input rate — Anthropic publishes 90% cache-hit discount on this tier.
Anthropic — Claude Sonnet 4.5 cachedInput Derived at 10% of input rate; same 90% cache-hit convention as Sonnet 4.6.
Anthropic — Claude Sonnet 4.5 batchInput Derived at 50% of standard input — Anthropic documents uniform 50% Batch discount.
Anthropic — Claude Sonnet 4.5 batchOutput Derived at 50% of standard output — Anthropic documents uniform 50% Batch discount.
Anthropic — Claude Haiku 4.5 cachedInput Derived at 10% of input rate — Anthropic 90% cache-hit discount convention.
OpenAI — GPT-5.4 Mini cachedInput Derived at 10% of input — OpenAI documents automatic 90% discount on cache hits across GPT-5.x tier.
OpenAI — GPT-5.4 Nano cachedInput Derived at 10% of input — OpenAI 90% cache-hit convention.
OpenAI — GPT-5.4 Nano batchInput Derived at 50% of input — OpenAI Batch API uniform 50% discount.
OpenAI — GPT-5.4 Nano batchOutput Derived at 50% of output — OpenAI Batch API uniform 50% discount.
OpenAI — GPT-5.4 Pro cachedInput Derived at 10% of input — OpenAI 90% cache-hit convention.
OpenAI — GPT-5.4 Pro batchInput Derived at 50% of input — OpenAI Batch API uniform 50% discount.
OpenAI — GPT-5.4 Pro batchOutput Derived at 50% of output — OpenAI Batch API uniform 50% discount.
OpenAI — GPT-5.2 cachedInput Derived at 10% of input; no residency uplift.
OpenAI — GPT-5.2 batchInput Derived at 50% of input.
OpenAI — GPT-5.2 batchOutput Derived at 50% of output.
OpenAI — GPT-5 cachedInput Derived at 10% of input.
OpenAI — GPT-5 batchInput Derived at 50% of input.
OpenAI — GPT-5 batchOutput Derived at 50% of output.
OpenAI — GPT-5.5 Pro cachedInput Derived at 10% of input — OpenAI does not publish a cached rate for *-pro models; using the family convention.
OpenAI — GPT-5.5 Pro batchInput Derived at 50% of input.
OpenAI — GPT-5.5 Pro batchOutput Derived at 50% of output.
OpenAI — GPT-5.2 Pro cachedInput Derived at 10% of input — pro-tier convention.
OpenAI — GPT-5.2 Pro batchInput Derived at 50% of input.
OpenAI — GPT-5.2 Pro batchOutput Derived at 50% of output.
OpenAI — GPT-5.1 batchInput Derived at 50% of input.
OpenAI — GPT-5.1 batchOutput Derived at 50% of output.
OpenAI — GPT-5 Pro batchInput Derived at 50% of input.
OpenAI — GPT-5 Pro batchOutput Derived at 50% of output.
OpenAI — GPT-5 Nano cachedInput Derived at 10% of input.
OpenAI — GPT-5 Nano batchInput Derived at 50% of input.
OpenAI — GPT-5 Nano batchOutput Derived at 50% of output.
Google — Gemini 3 Flash cachedInput Derived at 10% of input — Google caching discount convention ~90%.
Google — Gemini 3.1 Flash-Lite cachedInput Derived at 10% of input — Google caching convention.
Google — Gemini 3.1 Flash-Lite batchInput Derived at 50% of input — Google Batch API uniform 50% discount.
Google — Gemini 3.1 Flash-Lite batchOutput Derived at 50% of output — Google Batch API uniform 50% discount.
Google — Gemini 2.5 Pro cachedInput Derived at 10% of input.
Google — Gemini 2.5 Flash cachedInput Derived at 10% of input.
Google — Gemini 2.5 Flash-Lite cachedInput Derived at 10% of input — Google caching convention.
Google — Gemini 2.5 Flash-Lite batchInput Derived at 50% of input — Google Batch API uniform 50% discount.
Google — Gemini 2.5 Flash-Lite batchOutput Derived at 50% of output — Google Batch API uniform 50% discount.
Google — Gemini 2.0 Flash cachedInput Derived at 25% of input per Google 2.0 family caching rates.
Google — Gemini 2.0 Flash batchInput Derived at 50% of input — Google Batch API uniform 50% discount.
Google — Gemini 2.0 Flash batchOutput Derived at 50% of output — Google Batch API uniform 50% discount.
Google — Gemini 2.0 Flash-Lite cachedInput Derived at 10% of input — Google caching convention.
Google — Gemini 2.0 Flash-Lite batchInput Derived at 50% of input — Google Batch API uniform 50% discount.
Google — Gemini 2.0 Flash-Lite batchOutput Derived at 50% of output — Google Batch API uniform 50% discount.
xAI — Grok 4 (legacy) cachedInput Extrapolated at 25% of base.

Pricing is cross-verified against the LiteLLM community registry when available. Daily snapshots are kept in aicost_pricing_snapshots; every change is logged to aicost_price_changelog with old & new values for full audit trail. Read the full methodology →