Grok 4.5 Pricing: Plans, API Rates, and What It Costs
A plain-English reference to every published price for xAI's Grok 4.5 — the consumer tiers, the API rates, the enterprise routes, and how the total bill actually adds up.
Grok 4.5 is xAI's frontier model, priced at $2 per million input tokens and $6 per million output tokens on the xAI API, with a 500,000-token context window. Consumer access starts at $30 per month on SuperGrok, with full access on SuperGrok Heavy at $300 per month.
Vendor prices change without notice. Every rate on this page links back to xAI's own documentation so you can verify the current number before you commit a budget.
This guide covers the three ways teams actually buy Grok 4.5: through grok.com and X, through the xAI API, and through enterprise platforms like Databricks and Amazon Bedrock. It ends with worked cost examples and a like-for-like comparison to Claude Opus 5 and GPT-5.6.
Grok 4.5 API pricing at a glance
The xAI API charges $2.00 per million input tokens and $6.00 per million output tokens for Grok 4.5 on prompts under 200,000 tokens. Prompts at or above 200,000 tokens double to $4.00 input and $12.00 output per million.
Cached input is cheaper: $0.30 per million under 200K, $0.60 per million at 200K or more. Cache hits matter a lot for agent loops that keep replaying the same system prompt.
Priority tier processing bills at roughly 2x standard rates when the response confirms the tier. There is no published batch API discount for Grok 4.5, unlike the 20% batch discount on the older Grok 4.3.
Trying to model what Grok 4.5 will actually cost you at $2/$6 per million tokens once tool calls and priority tier get added on? We can price it against your current traffic pattern in one working session.
Book a Grok 4.5 cost auditContext window, tool calls, and the small charges people miss
Grok 4.5 ships with a 500,000-token context window on the xAI API. That is big enough for a mid-sized codebase, a stack of contracts, or a full multi-turn agent trace inside one call.
Tool calls bill separately from tokens. Web search and X search each cost $5.00 per 1,000 calls. Code execution runs $5.00 per 1,000 calls. Collections search costs $2.50 per 1,000 calls. Image and video understanding stay token-based rather than per-call.
These small line items add up fastest on agent workloads. An agent that runs three tool calls per turn across 10,000 sessions is a $150 line item on top of the token bill.
Consumer plans: grok.com, X Premium+, and SuperGrok Heavy
Consumer access to Grok 4.5 comes through three plans. SuperGrok is $30 per month or roughly $300 per year. X Premium+ is $40 per month and bundles Grok with X platform perks. SuperGrok Heavy is $300 per month and is the tier that ships the biggest Grok 4.5 usage quotas.
The free grok.com tier gives limited Grok access with lower rate caps and staged access to the newest model. Individual professionals doing deep research or long agentic sessions usually need SuperGrok or higher.
For teams, the consumer plans are almost never the right buy. They lack shared billing, admin controls, and audit logs. Teams should default to the xAI API or an enterprise platform.
Enterprise access: xAI API, Databricks, and Bedrock
Most business buyers reach Grok 4.5 through the xAI API on a pay-as-you-go account, with rate-limit tiers that unlock automatically as cumulative spend since January 1, 2026 grows. Tiers never downgrade once earned.
xAI also distributes select Grok models through Databricks and Amazon Bedrock. Pricing on those platforms adds a marketplace margin on top of the raw token cost, but buyers get consolidated billing, private networking, and their existing data-governance controls.
Verify current enterprise pricing inside your Databricks or AWS console before committing. Marketplace listings are updated on a different cadence from xAI's own docs.
In our own work at Layer3Labs — running the /keyword-gap and /mindmap-pass content routines across dozens of AI-content sites — the pattern we see with every new frontier launch is that the first published API price is stable enough to plan against for a quarter, but consumer-plan quotas move week to week during rollout. Budget the API rate, not the SuperGrok cap.
Worked cost examples: what 1,000 requests actually costs
A short chat request of 1,500 input tokens and 500 output tokens costs about $0.006 per call on standard-tier Grok 4.5 API. One thousand of those requests runs roughly $6.00 in tokens.
A retrieval-augmented answer with 15,000 input tokens (context + system prompt) and 1,000 output tokens costs about $0.036 per call. One thousand of those calls runs about $36.00 — plus any tool-call fees if the agent searches the web.
A long-context code review at 180,000 input tokens and 4,000 output tokens costs about $0.384 per call at standard rates. Cross the 200K threshold and the same call jumps to about $0.768 because both input and output rates double.
These are token math only. Add tool calls at $5 per 1,000 for web or code execution, and add priority-tier surcharges if latency guarantees matter to you.
Grok 4.5 vs. Claude Opus 5 and GPT-5.6 pricing
Grok 4.5 sits in the cheaper half of frontier-model pricing. At $2 input and $6 output per million tokens, it undercuts Claude Opus 5 by a wide margin at the input end and lands close to GPT-5.6's mid-tier reasoning pricing on output.
Compare on the workload, not the sticker. A prompt-heavy workload with modest generation (RAG, classification, extraction) favors Grok 4.5's low input rate. A generation-heavy workload (long-form writing, deep reasoning traces) narrows the gap because output tokens dominate.
Always run a small pilot before switching. Vendor benchmarks rarely match your prompt distribution, and Claude and GPT often win or lose on tool-use reliability or refusal behavior rather than raw token price. Cached input, batch discounts, and volume tiers can flip the ranking again at scale.
How to choose the right Grok 4.5 buying route
Pick the API for anything programmatic. Even one production workflow is cheaper and more controllable on the API than on a stack of SuperGrok seats.
Pick SuperGrok Heavy for individual heavy users who live in the chat UI — research analysts, technical founders, solo consultants. It is the only consumer tier that reliably gives full Grok 4.5 access at high quotas today.
Pick Databricks or Bedrock when procurement or data-residency rules make xAI direct-billing hard. Accept the marketplace margin in exchange for one throat to choke and existing governance.
Whatever route you pick, revisit the pricing page every quarter. Frontier-model rates are a moving target and xAI has cut them before.
Frequently Asked Questions
- Grok 4.5 costs $2.00 per million input tokens and $6.00 per million output tokens on the xAI API for prompts under 200,000 tokens. Prompts at or above 200,000 tokens double to $4.00 input and $12.00 output per million. Cached input runs $0.30 per million below the threshold. Verify current rates at docs.x.ai/developers/pricing before committing.
- Grok 4.5 supports a 500,000-token context window on the xAI API. Prompts crossing the 200,000-token mark bill at the higher long-context rate, so plan for that jump if you routinely load large codebases or document sets.
- SuperGrok is $30 per month or about $300 per year. X Premium+ is $40 per month and bundles Grok with X platform features. SuperGrok Heavy is $300 per month and gives the highest Grok 4.5 quotas on the consumer side.
- Grok 4.5 undercuts Claude Opus 5 by a wide margin at the input rate and lands close to GPT-5.6 mid-tier reasoning pricing on output. The actual win depends on your input-to-output ratio. Prompt-heavy workloads favor Grok 4.5. Verify each vendor's current pricing page before choosing.
- No published batch API discount exists for Grok 4.5, unlike the 20% batch discount on the older Grok 4.3. If batch pricing matters to your workload, confirm the current policy at docs.x.ai/developers/pricing.
- Web search and X search each cost $5.00 per 1,000 calls on the xAI API. Code execution costs $5.00 per 1,000 calls. Collections search costs $2.50 per 1,000 calls. Image and video understanding stay token-based rather than per-call. These charges are added on top of the token bill.
- The free tier on grok.com gives limited Grok access with lower rate caps and staged model access. Serious daily use requires SuperGrok, X Premium+, or higher. For production workloads, the xAI API's pay-as-you-go pricing is usually the better path.
Not sure whether Grok 4.5 belongs in your stack?
We map Grok 4.5's per-token math and rate-limit tiers to your actual workloads, then tell you whether it beats what you already run.
Book a Grok 4.5 cost audit