OpenRouter model IDs — how do I find the right format?

OpenRouter model IDs follow provider/model-name: openai/gpt-4o, anthropic/claude-3-5-sonnet, meta-llama/llama-3.3-70b-instruct, google/gemini-flash-1.5, mistralai/mixtral-8x7b-instruct. Browse the full list at openrouter.ai/models. Add :free suffix for free models only: meta-llama/llama-3.3-70b-instruct:free. Add :nitro suffix for lower-latency routing to the fastest available provider.

OpenRouter vs direct API — when should I use OpenRouter?

Use OpenRouter when: (1) you want to switch between GPT-4, Claude, and LLaMA with one API key; (2) you want automatic failover when a provider is down; (3) you want to compare model costs and outputs; (4) you're prototyping and don't want separate accounts for each provider. Use the direct API when: (1) you need provider-specific features not exposed through OpenRouter; (2) latency is critical (OpenRouter adds 20-50ms routing overhead); (3) you need provider-native rate limits.

Free 4 min read

OpenRouter Not Working?

OpenRouter API 401, model routing error, credits exhausted (402), provider error surfaced through routing, or streaming failing? Check live status and fix it fast.

OpenRouter — live status

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

Full status →

OpenRouter routes to multiple providers

If a specific provider (Anthropic, OpenAI, etc.) is degraded, only their models are affected. Other providers' models continue to work. Check which provider the failing model uses at openrouter.ai/models.

What's wrong? Diagnose fast

🔑

API 401 — invalid key

OpenRouter keys start with sk-or-. Generate at openrouter.ai/keys. Header: Authorization: Bearer sk-or-YOUR_KEY. Optional recommended headers: HTTP-Referer (your site) and X-Title (your app name) for usage tracking.

🔍

Model not found (404)

Use provider/model-name format: openai/gpt-4o, anthropic/claude-3-5-sonnet, meta-llama/llama-3.3-70b-instruct. Browse exact IDs at openrouter.ai/models. Add :free for free-only routing. Add :nitro for fastest provider routing.

💸

Credits exhausted (402)

Check balance at openrouter.ai/credits. Free: $1 initial one-time credit. Add more at openrouter.ai/credits (Stripe or crypto). Each call costs model price + 5% OpenRouter fee. Request log: openrouter.ai/activity.

⚠

Provider error in response

Check error.metadata.provider_name (which provider failed) and error.metadata.raw (original error). Provider 529 = overloaded. Fix: add route=fallback parameter to try other providers, or use :nitro model suffix. Switch to a different model from another provider.

📡

Streaming fails / cuts off

Stream: true works the same as OpenAI. If stream cuts off: some proxies buffer SSE. Test with direct cURL. For long-running streams, some providers have different timeout limits than the default. Use transforms=[] to disable OpenRouter post-processing on the stream.

📊

Wrong model was used

OpenRouter may route to a different variant than expected. Check x-openrouter-model response header to see the actual model used. Add provider preference: route=lowest-cost or route=nitro. Specify providers[] array to restrict routing to specific providers.

OpenRouter API quick reference

curl (OpenAI-compatible)

curl -X POST "https://openrouter.ai/api/v1/chat/completions" \
  -H "Authorization: Bearer sk-or-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -H "HTTP-Referer: https://your-site.com" \
  -H "X-Title: Your App Name" \
  -d '{
    "model": "anthropic/claude-3-5-sonnet",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'
# Check x-openrouter-model header in response for actual model used

Python (OpenAI SDK drop-in)

from openai import OpenAI

client = OpenAI(
    api_key="sk-or-YOUR_KEY",
    base_url="https://openrouter.ai/api/v1",
    default_headers={
        "HTTP-Referer": "https://your-site.com",
        "X-Title": "Your App Name",
    },
)

response = client.chat.completions.create(
    model="openai/gpt-4o",
    messages=[{"role": "user", "content": "Hello!"}],
)

Provider routing with fallback

{
  "model": "meta-llama/llama-3.3-70b-instruct",
  "messages": [...],
  "route": "fallback",        // try other providers if primary is down
  "providers": {
    "order": ["Together", "Fireworks"],  // prefer these providers
    "allow_fallbacks": true
  }
}

OpenRouter model suffixes

Suffix	Example	Effect
(none)	openai/gpt-4o	Default routing — lowest latency available
:free	meta-llama/llama-3.3-70b-instruct:free	Only route to free providers — $0 cost
:nitro	openai/gpt-4o:nitro	Fastest provider routing (may cost more)
:floor	anthropic/claude-3-5-sonnet:floor	Cheapest provider routing
:online	openai/gpt-4o:online	Adds live web search to response

Step-by-step fix

1

Check live OpenRouter status

Visit prismix.dev/service/openrouter. Remember: only the specific provider's models are affected when a provider (not OpenRouter itself) is degraded.
2

Fix API 401 errors

Your key must start with sk-or-. Generate at openrouter.ai/keys. Header: Authorization: Bearer sk-or-YOUR_KEY. OpenAI SDK: set base_url="https://openrouter.ai/api/v1". Add HTTP-Referer and X-Title headers for proper usage tracking in your dashboard.
3

Fix model not found

Use the exact provider/model-name format from openrouter.ai/models. Common models: openai/gpt-4o, anthropic/claude-3-5-sonnet, meta-llama/llama-3.3-70b-instruct. For free usage: append :free suffix.
4

Fix credits exhausted (402)

Check your balance at openrouter.ai/credits. Each call costs the model's base price plus OpenRouter's 5% fee. Add credits via Stripe or crypto. View your request history and costs at openrouter.ai/activity. To use completely free models: use the :free model suffix.
5

Fix provider-specific errors

Provider errors are surfaced in the response: error.metadata.provider_name shows which provider failed, error.metadata.raw shows the raw error. Fix: add "route": "fallback" to your request body to automatically try alternative providers.

🔔

Get alerted when OpenRouter goes down

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