xAI API pricing

Billing rules Veqen applies

• Token models: cached input at the cache-read rate, cache writes at the write rate; batch and service tiers at their own rates where published; long-context rates once the full prompt passes the threshold.
• Realtime models: audio tokens at their own input and output rates.
• Image models: per image, with each size and quality priced separately. When the API reports image tokens, those are the bill.
• Video models: per second of generated video, at the requested resolution's rate.
• Audio models: per second of audio or per character of text, or per token where the model bills tokens.

Also recognized as: grok, x_ai (framework, gateway and OpenTelemetry names).

Reconcile your bill

xAI doesn't offer a cost API, so Veqen's ledger is your running record of spend. Enter each month's bill to reconcile it and book it in the close. How reconciliation works.

Prices come from xAI's pricing page. Negotiated rates or committed-use discounts can be set as custom prices.