xAI

The combined company listed on Nasdaq in June 2026. Positioning is a full-stack frontier lab: one flagship text/coding line (Grok), plus first-party image/video (Grok Imagine) and realtime voice APIs, with unusually deep native integration to X (X Search) — all served proprietary via the xAI API, Grok Build (its coding agent/CLI), Google Cloud Vertex AI, Microsoft Foundry, and gateways such as OpenRouter, Vercel and Cloudflare.

A Access and limits

Sign up at console.x.ai, load prepaid credits, then create a key on the API Keys page and pass it as Authorization: Bearer $XAI_API_KEY. No free tier or free credit grant is documented — usage is prepaid credits or monthly invoiced billing above a self-set spending limit; Indian payment cards are not supported. Base URL https://api.x.ai/v1 is OpenAI-SDK compatible (Responses API is recommended; Chat Completions is documented as legacy/deprecated); a native gRPC xai-sdk and a Vercel AI SDK provider also exist.

Rate-limit tiers are set by cumulative API spend since 2026-01-01: Tier 0 ($0, default), Tier 1 ($50), Tier 2 ($250), Tier 3 ($1,000), Tier 4 ($5,000), Enterprise on request; tiers never downgrade and apply to text models only (Voice/Imagine increases via sales@x.ai). Example limits: grok-4.6 150 RPS / 50M TPM at Tier 0 rising to 500 RPS / 100M TPM at Tier 4; grok-4.3 and grok-4.20 37 RPS / 10M TPM at Tier 0. Clusters are us-east-1 and us-west-2 (voice, STT and TTS are us-east-1 only); Grok 4.5 is additionally available to EU console users.

IMPORTANT PRICING STRUCTURE: every text model has two-tier long-context pricing — once a request's prompt reaches 200K tokens, ALL tokens in that request bill at the long-context rate (exactly 2x the base rate on input, cached input and output for every model). The prices recorded per model here are the BASE (<200K prompt) tier. Batch API (~24h turnaround, exempt from rate limits) gives a 20% discount on grok-4.3, grok-4.20-0309-reasoning, grok-4.20-0309-non-reasoning and grok-4.20-multi-agent-0309; other models get no batch discount, and grok-4.6, grok-4.5, grok-build-0.1 and grok-imagine-image-quality do not accept batch at all.

Priority Processing (service_tier: "priority") bills at 2x standard rates on all token types, applied after cache discounts, and only when the response confirms the priority tier. Server-side tools bill on top of tokens: web_search $5/1k calls, x_search $5/1k, code_execution $5/1k, attachment_search $10/1k, collections_search $2.50/1k; view_image, view_x_video and remote MCP tools are token-billed only. Storage: files $0.025/GiB/day, collections $0.10/GiB/day, downloads $0.20/GiB.

A $0.05 per-request fee applies to usage-guideline violations caught pre-generation on the Responses API. Retirements: eight legacy slugs (grok-4-1-fast-reasoning/non-reasoning, grok-4-fast-reasoning/non-reasoning, grok-4-0709, grok-code-fast-1, grok-3, grok-imagine-image-pro) were retired 2026-05-15 and now silently redirect (to grok-4.3, grok-build-0.1, or grok-imagine-image-quality) and bill at the target's price; grok-imagine-image-quality retires 2026-11-02 and will redirect to grok-imagine-image-2.0 at quality=low. Caveat: xAI's release notes contain no entry for grok-4.3, so its release date below comes from press coverage rather than an official xAI page; every price and context window below was read from docs.x.ai.

C Calling the API

xAI accepts requests in the OpenAI Chat Completions format, so any OpenAI-compatible client works by changing the base URL. Example uses grok-build-0.1.

curl https://api.x.ai/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-build-0.1",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

from openai import OpenAI

client = OpenAI(
    base_url="https://api.x.ai/v1",
    api_key=os.environ["API_KEY"],
)

response = client.chat.completions.create(
    model="grok-build-0.1",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

A compatibility layer is not always the provider's full API — caching, strict tool schemas and structured output are commonly unsupported there. Check the official docs before relying on an advanced feature.

D Official references

• Homepage
• API documentation
• Pricing