Comparisons

Grok 4.5 vs Grok 4.6

On provider list prices, Grok 4.5 costs $2 per million input tokens against $2 for Grok 4.6: effectively level. Output is $6 against $6.

Grok 4.5 Grok 4.6
LabSpaceXAISpaceXAI
AccessAPI onlyAPI only
Context window500K tokens500K tokens
List price, input$2 / M tokens$2 / M tokens
List price, output$6 / M tokens$6 / M tokens
Cached input$0.3 / M tokens$0.5 / M tokens
LicenseProprietary APIProprietary API
Fine-tunableNoNo

Specifications and provider list prices from the Allocate catalog, checked 2026-07-21.

What the numbers say

Take 1,000,000 requests a month at 1,200 input and 350 output tokens each. That workload costs $4,500 a month on Grok 4.5 and $4,500 on Grok 4.6 at list: a gap of $0.

Grok 4.5$2$6
Grok 4.6$2$6
InputOutput

Choose Grok 4.5 for

  • Frontier serving with no weights to manage
Grok 4.5 details

Choose Grok 4.6 for

  • Frontier serving with no weights to manage
Grok 4.6 details

Common questions

Which is cheaper, Grok 4.5 or Grok 4.6?

Grok 4.5, on this workload shape. At list prices it is $2/$6 per million tokens in and out against $2/$6 for Grok 4.6. Billed on Allocate: $2.14/$6.42 against $2.14/$6.42.

Which has the bigger context window?

They match: both read 500,000 tokens (500K) per request.

Can I fine-tune Grok 4.5 or Grok 4.6?

No. Both are closed models served over API. If you want a model you can train and own, start from an open-weights base in the catalog.

Related comparisons

Run the numbers on your workload

Or don’t choose. On Allocate a route name is the contract: point yours at one model today, swap to the other tomorrow, and compare them on your live traffic with per-token metering.