ChangeRadar
Model API

xAI · Grok

xAI · source page ↗ · last checked Oct 9, 2026, 6:00 AM

Current pricing

● live captured Oct 9, 2026, 6:00 AM
Model / itemInputOutput
grok-4.3$1.25 / 1M tokens$2.50 / 1M tokens · Chat API; 1M context. Cached input: $0.20 / 1M tokens
grok-4.20-multi-agent-0309$1.25 / 1M tokens$2.50 / 1M tokens · Chat API; 1M context. Cached input: $0.20 / 1M tokens
grok-4.20-0309-reasoning$1.25 / 1M tokens$2.50 / 1M tokens · Chat API; 1M context. Cached input: $0.20 / 1M tokens
grok-4.20-0309-non-reasoning$1.25 / 1M tokens$2.50 / 1M tokens · Chat API; 1M context. Cached input: $0.20 / 1M tokens
grok-build-0.1$1.00 / 1M tokens$2.00 / 1M tokens · Chat API; 256k context. Cached input: $0.20 / 1M tokens
grok-imagine-image$0.002 / image (media input)$0.02 / image (1K and 2K resolution) · Image generation (Text, Image to Image)
grok-imagine-image-quality$0.01 / image (media input)$0.05 / image (1K); $0.07 / image (2K) · Image generation (Text, Image to Image)
grok-imagine-video$0.01 / sec + $0.002 / image (media input)$0.05 / sec (480p); $0.07 / sec (720p) · Video generation (Text, Image, Video to Video)
grok-imagine-video-1.5-preview$0.01 / image (media input)$0.08 / sec (480p); $0.14 / sec (720p) · Video generation (Image to Video)
Voice — Realtime$0.05 / min ($3.00 / hr) · Voice Agent API
Voice — Realtime Text Input$0.004 / message · Voice Agent API
Voice — Text to Speech$15.00 / 1M characters
Voice — Speech to Text$0.10 / hr (REST); $0.20 / hr (Streaming)
Tool — Web Search$5 / 1k calls · Server-side tool invocation (web_search); billed on top of token usage
Tool — X Search$5 / 1k calls · Server-side tool invocation (x_search); billed on top of token usage
Tool — Code Execution$5 / 1k calls · Server-side tool invocation (code_execution / code_interpreter)
Tool — File Attachments$10 / 1k calls · Server-side tool invocation (attachment_search)
Tool — Collections Search (RAG)$2.50 / 1k calls · Server-side tool invocation (collections_search / file_search)
File storage$0.025 / GiB / day · Files and collections pricing
Collection storage$0.10 / GiB / day · Files and collections pricing
File / Collection downloads$0.20 / GiB downloaded · Flat rate per GiB transferred
Grok$2.00 / 1M tokens$6.00 / 1M tokens
VisionStarting at $0.02 / image—
Real-time AudioStarting at $0.02 / sec—

Usage/token API pricing (no subscription tiers). All prices in USD. IMPORTANT: the requested page itself (https://docs.x.ai/docs/models) currently contains NO price table — it only carries model-selection guidance, model aliases, image-input limits, and knowledge-cutoff notes (page last updated May 29, 2026). The verified prices above were extracted from the companion canonical pricing page https://docs.x.ai/docs/pricing (last updated May 27, 2026), which explicitly states "for per-model details, see the models page" — they are the same xAI docs set. Note: an automated WebFetch summary of the /docs/models URL fabricated a pricing table that did not exist on that page, so figures were re-verified against the raw /docs/pricing scrape. Batch API: text/language models billed at 20%-50% off standard rates (image/video generation billed at standard rates in Batch). Image Understanding, X Video Understanding, and Remote MCP tools are token-based (no flat invocation fee). Prices subject to change.

Rate & usage limits ● live

Tier Tokens/min Notes
Tier 0 10M, 15M, 25M, 45M, 85M (model-dependent) $0 spend threshold (default). RPS per model: 37 RPS (grok-4.20, grok-4.3, grok-build), 150 RPS (grok-4.5/4.6/4.7), 9 RPS (grok-4.20-multi-agent), 6 RPS (image), 10 RPS (video), 50 RPS (TTS), 10 RPS (STT)
Tier 1 15M, 15M, 25M, 45M, 85M (model-dependent) $50 cumulative spend threshold. RPS per model: 50 RPS (grok-4.20, grok-4.3, grok-build), 172 RPS (grok-4.5/4.6/4.7), 12 RPS (grok-4.20-multi-agent), 12 RPS (image), 20 RPS (video), 50 RPS (TTS), 10 RPS (STT)
Tier 2 25M, 25M, 25M, 45M, 85M (model-dependent) $250 cumulative spend threshold. RPS per model: 75 RPS (grok-4.20, grok-4.3, grok-build), 208 RPS (grok-4.5/4.6/4.7), 18 RPS (grok-4.20-multi-agent), 25 RPS (image), 39 RPS (video), 100 RPS (TTS), 20 RPS (STT)
Tier 3 45M, 45M, 45M, 45M, 85M (model-dependent) $1,000 cumulative spend threshold. RPS per model: 125 RPS (grok-4.20, grok-4.3, grok-build), 312 RPS (grok-4.5/4.6/4.7), 31 RPS (grok-4.20-multi-agent), 50 RPS (image), 79 RPS (video), 250 RPS (TTS), 30 RPS (STT)
Tier 4 85M (all language models) $5,000 cumulative spend threshold. RPS per model: 208 RPS (grok-4.20, grok-4.3, grok-build), 500 RPS (grok-4.5/4.6/4.7), 56 RPS (grok-4.20-multi-agent), 100 RPS (image), 158 RPS (video), 500 RPS (TTS), 40 RPS (STT)
Enterprise — Available on request. Contact sales@x.ai for enterprise-grade capacity.

Limits are per-second (RPS) and tokens-per-minute (TPM) per model. RPM is derived from RPS (RPM = RPS × 60, with minimum of 2). No per-day limits published. Limits scale with tier based on cumulative API spend since January 1, 2026. Dashboard-personalized limits available in xAI Console. Image and video models use RPS only. Voice models use RPS and concurrent sessions. Per-request rate limit errors return HTTP 429.

Source: https://docs.x.ai/developers/rate-limits

In the news · xAI

Importance-filtered press coverage (Google News) mentioning xAI. Headlines link to the original; verify before acting.

Change history