Grok 4.6 API Pricing & Cost Calculator

By TechCompare · Updated · How we check

Grok 4.6 at $2/$6 is priced like Qwen 3.8 Max, but xAI's 200,000-token length tier is the catch: one long call bills entirely at $4/$12. Keep prompts under 200,000 tokens to stay on the base rate.

A newer version of Grok 4.6 is out: Grok 4.7. See Grok 4.7 pricing or compare every Grok model.

Calculator

Cost Comparison

Based on 100,000 input tokens (50% cached), 5,000 output tokens, and 100 requests.Prices come live from OpenRouter at each maker's own price where it's listed. Temporary deals are left out.

Loading model prices

Grok 4.6 is xAI's follow-up to Grok 4.5, billing $2/$6 without raising Grok 4.5's $2/$6 token rate but introducing a prompt-length tier: once a prompt reaches 200,000 tokens, the whole request moves to $4/$12, including cache reads at $1.

SpaceXAI: Grok 4.6

xAI's own price

Input
$2
Output
$6
Cached
$0.50

per 1M tokens

Batch available

Price checked 2026-11-19

How this is calculated

Grok 4.6 is priced at $2.00 per million input tokens and $6.00 per million output tokens for prompts under 200,000 tokens, and $4.00/$12.00 for prompts at or above that. Cache reads are $0.50/M below the threshold and $1.00/M above. Batch takes 20% off both rows ($1.60/$4.80), and the full context window is 500K.

Verdict

A 100K + 5K call costs $0.23, matching Qwen 3.8 Max exactly at the low tier. Cross the 200,000-token prompt threshold and the whole request bills at $4/$12, so a 250K + 5K call runs $1.06 versus $0.53 on a tierless $2/$6 model.

Cache reads at $0.50/M (below the threshold) are weaker than Alibaba's $0.25 and DeepSeek's $0.044, so stable-prompt loops cost more on xAI than on either. Qwen has the larger context window at 1M versus Grok's 500K. Requests whose total context exceeds 500K cannot fit Grok's limit.

More API Standalones scenarios

Frequently asked questions

How does prompt-length billing work on Grok 4.6?
If a request's prompt is below 200,000 tokens it bills at $2/M input and $6/M output. At or above 200,000 the entire request bills at $4/M input and $12/M output, including cache reads which go from $0.50/M to $1.00/M. There's no partial split: one long call pays the higher rate for all of it.
How does Grok 4.6 compare to Qwen 3.8 Max?
Same $2/$6 headline rate, but Qwen's cache reads are $0.25/M versus Grok's $0.50/M. Qwen has the context advantage at 1M versus Grok's 500K. Consider Grok when xAI's real-time integrations fit your workload. On pure token cost, Qwen wins under 200,000-token prompts.
How much does Grok 4.6 cost per million output tokens?
$6.00 per million below 200,000-token prompts, $12.00 per million at or above. Batch brings output to $4.80 per million. Cache reads are $0.50 per million in the low tier and $1.00 in the high tier.