xAI Grok API 401 — how to fix?

xAI API base URL: https://api.x.ai/v1 (note: x.ai not openai.com). Key format: xai-YOUR_KEY (starts with 'xai-'). Auth: Authorization: Bearer xai-YOUR_KEY. Generate at console.x.ai. OpenAI SDK: from openai import OpenAI; client = OpenAI(base_url='https://api.x.ai/v1', api_key='xai-YOUR_KEY'). This is OpenAI-API compatible.

What are the exact Grok model names?

xAI model IDs (exact): grok-3-beta (most capable, 131k context), grok-3-mini-beta (fast and cheap, 131k context), grok-2-vision-1212 (image understanding), grok-2-1212 (stable release). Do NOT use 'grok', 'grok-1', 'grok-3', or 'grok-beta' — these will 404. Check https://api.x.ai/v1/models for the latest list.

Grok chat on X (Twitter) not working — how to fix?

Grok on X requires X Premium or Premium+ subscription ($8-16/month). Verify at x.com/i/premium. Grok image generation requires Premium+ (higher tier). Grok is accessible at grok.com (separate from X app, sometimes faster during peak hours). If chat is stuck: clear browser cache, try grok.com instead of X app, check status.x.ai for platform outages.

xAI 4 min read

Grok Not Working?

xAI API 401 (wrong base URL — use api.x.ai/v1, not api.openai.com), model not found (grok-3-beta, not grok), Grok chat on X needs Premium subscription, or 503 outage? Fix it fast.

xAI Grok — live status

Updated every 5 minutes. Full history at prismix.dev/service/xai.

Full status →

What's wrong? Diagnose fast

🔑

API 401 — wrong base URL

Most common mistake: using OpenAI's URL with Grok. The correct xAI base URL is https://api.x.ai/v1 (NOT https://api.openai.com/v1). Key starts with "xai-", generated at console.x.ai. Auth: Authorization: Bearer xai-YOUR_KEY. OpenAI SDK: set base_url="https://api.x.ai/v1".

💬

Model not found — wrong IDs

Exact model IDs: grok-3-beta (most capable, 131k ctx), grok-3-mini-beta (fast, cheap), grok-2-vision-1212 (vision), grok-2-1212 (stable). NOT valid: "grok", "grok-1", "grok-3", "grok-beta". Get current list: GET https://api.x.ai/v1/models.

⚡

Rate limit 429

Free tier (early access): 25 requests/hour, 60,000 tokens/hour. On 429: check x-ratelimit-remaining and x-ratelimit-reset headers, implement exponential backoff (5s start). Paid tier available at console.x.ai for higher limits.

🐦

Grok chat on X not responding

Grok on X (Twitter) requires X Premium or Premium+ subscription ($8-16/mo). Verify at x.com/i/premium. Grok image generation needs Premium+. Alternative: use grok.com directly (works the same, sometimes faster). Check status.x.ai for platform-wide outages.

📏

Context limit exceeded

Grok-3-beta and grok-3-mini-beta: 131,072 token context (prompt + completion combined). Grok-2-1212: 131,072 tokens. For conversations over 100k tokens: summarize history, truncate old messages. If you hit context limits with the default models, this is the maximum — no larger context variant available.

🖼

Vision not working

Vision requires grok-2-vision-1212 model specifically. Pass image as base64 or URL in messages array: {"type": "image_url", "image_url": {"url": "https://..."}}. Grok-3-beta does NOT support vision as of 2025 — use grok-2-vision-1212 for image input.

xAI Grok API quick reference

OpenAI SDK (drop-in — just change base_url)

from openai import OpenAI

client = OpenAI(
    base_url="https://api.x.ai/v1",   # NOT api.openai.com
    api_key="xai-YOUR_KEY",            # key starts with xai-
)

response = client.chat.completions.create(
    model="grok-3-beta",               # NOT "grok" or "grok-3"
    messages=[{"role": "user", "content": "What is 2+2?"}],
    max_tokens=1024,
)
print(response.choices[0].message.content)

Vision (image understanding)

response = client.chat.completions.create(
    model="grok-2-vision-1212",   # use THIS model for vision, not grok-3-beta
    messages=[{
        "role": "user",
        "content": [
            {"type": "text", "text": "What's in this image?"},
            {"type": "image_url", "image_url": {"url": "https://example.com/img.jpg"}}
        ]
    }],
)

curl — verify authentication

curl https://api.x.ai/v1/chat/completions \
  -H "Authorization: Bearer xai-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-3-mini-beta",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

# List available models:
curl https://api.x.ai/v1/models -H "Authorization: Bearer xai-YOUR_KEY"

Grok models quick reference

Model ID	Context	Capabilities
grok-3-beta	131,072 tokens	Best quality, text only
grok-3-mini-beta	131,072 tokens	Fast + cheap, text only, reasoning
grok-2-vision-1212	131,072 tokens	Text + image understanding
grok-2-1212	131,072 tokens	Stable release, GPT-4-level quality

Step-by-step fix

1

Check live xAI status

Visit prismix.dev/service/xai. Also status.x.ai. xAI API and grok.com have independent uptime from the X (Twitter) platform.
2

Fix API 401 — correct base URL

Base URL: https://api.x.ai/v1 (NOT api.openai.com). Key: xai-YOUR_KEY. Auth: Authorization: Bearer xai-YOUR_KEY. Generate at console.x.ai.
3

Fix model not found

Use grok-3-beta or grok-3-mini-beta. For vision: grok-2-vision-1212. List all: GET /v1/models.
4

Fix rate limit 429

Free tier: 25 req/hr. Check x-ratelimit-reset header for when it resets. Exponential backoff: start at 5s. Upgrade at console.x.ai for higher limits.
5

Fix Grok on X (chat)

Requires X Premium or Premium+ subscription. Verify at x.com/i/premium. Alternatively use grok.com directly — same features, independent uptime from X app.

🔔

Get alerted when xAI Grok goes down

Star xAI on Prismix and get emailed the moment status changes. Free, no credit card.