Most developers using Claude Code still hit Anthropic’s API directly. That instinct costs you money and locks you into a single provider’s pricing and availability. The bigger story here isn’t just saving a few cents per million tokens — it’s about a structural shift in how AI models are accessed.
The old model was simple: pick one lab, one API key, one bill. The new model is routing. Platforms like OpenRouter aggregate dozens of models from multiple providers under a single endpoint. You get competitive pricing, automatic fallback, and the ability to switch models without rewriting code. For Claude Code specifically, this means you can use Anthropic’s models through a marketplace that often undercuts direct pricing while adding redundancy.
OpenRouter routes your Claude Code requests to the cheapest available provider automatically.
Here’s how to set it up, step by step.
Step 1: Understand the Cost Advantage
Direct Anthropic API pricing for Claude 3.5 Sonnet is $3 per million input tokens and $15 per million output tokens. OpenRouter’s pricing for the same model typically sits around $2.80–$3.00 input and $14.50–$15.00 output — a small but consistent discount. The real savings come from two things: you can cap spending per session (avoid runaway costs), and you can route to cheaper models (like Haiku or Llama) when your task doesn’t need Sonnet’s power.
Why this matters beyond the pennies: OpenRouter lets you treat models as interchangeable commodities. If Anthropic raises prices or suffers an outage, you can shift traffic to another provider serving the same model within minutes. That’s not a feature — it’s a hedge against vendor lock-in.
Step 2: Create an OpenRouter Account
- Go to openrouter.ai and sign up with your email or Google account.
- Navigate to the API Keys section and generate a new key. Give it a label like “Claude Code” so you remember its purpose.
- Set a spending cap — start at $25. This cap matters more here than in a normal chat because a code agent running on a schedule can burn through tokens overnight with nobody watching.
Expected outcome: You now have an OpenRouter API key that works with any model they support, including all Claude variants.
Step 3: Configure Claude Code to Use OpenRouter
Claude Code (the terminal-based agent from Anthropic) supports custom API endpoints via environment variables. OpenRouter exposes a standard OpenAI-compatible endpoint, so the configuration is straightforward.
- Open your terminal and set the following environment variables:
“`
export ANTHROPIC_BASE_URL=https://openrouter.ai/api/v1
export ANTHROPIC_API_KEY=sk-or-v1-your-key-here
“`
Replace the key with the one you generated.
- Optionally, set a default model:
“`
export CLAUDE_MODEL=anthropic/claude-3.5-sonnet
“`
OpenRouter uses the format provider/model-namecodecode for model identifiers.
- Start Claude Code as usual. It will now route all requests through OpenRouter.
Expected outcome: Claude Code runs identically, but usage is billed through OpenRouter at their rates. You can verify by checking your OpenRouter dashboard for live token counts.
Step 4: Add Fallback Models for Reliability
One of OpenRouter’s strongest features is automatic fallback. If the primary model is down or rate-limited, you can define a secondary model to take over.
- In your environment, set a fallback by appending it to the model parameter:
“`
export CLAUDE_MODEL=anthropic/claude-3.5-sonnet:google/gemini-1.5-pro
“`
The colon separates the primary from the fallback.
- Or use OpenRouter’s route configuration in your code. For example, in a Python script:
“`python
import openai
client = openai.OpenAI(
base_url=”https://openrouter.ai/api/v1″,
api_key=”sk-or-v1-your-key-here”
)
response = client.chat.completions.create(
model=”anthropic/claude-3.5-sonnet”,
fallback_models=[“google/gemini-1.5-pro”, “meta-llama/llama-3-70b”],
messages=[{“role”: “user”, “content”: “Write a function to sort a list.”}]
)
“`
Expected outcome: If Claude is unavailable, the request automatically routes to Gemini or Llama without throwing an error. Your code keeps running.
Step 5: Compare Providers Side by Side
Not all routes are equal. The table below shows typical pricing and latency for the same Claude model through different providers on OpenRouter. Prices are approximate and change — always check the dashboard.
| Provider | Input Cost (per MTok) | Output Cost (per MTok) | Avg. Latency | Notes |
|---|---|---|---|---|
| Anthropic (direct) | $3.00 | $15.00 | ~1.5s | No markup, but no fallback |
| OpenRouter (via Anthropic) | $2.90 | $14.80 | ~1.8s | Slight discount, fallback available |
| OpenRouter (via DeepSeek) | $0.14 | $0.28 | ~2.5s | 20x cheaper, but different model behavior |
| OpenRouter (via Google) | $1.25 | $5.00 | ~1.2s | Gemini 1.5 Pro, good for structured tasks |
Tradeoff to watch: Cheaper providers often have higher latency or different output patterns. Test your specific use case before routing everything to the cheapest option.
Step 6: Automate Spending Controls
OpenRouter lets you set per-key spending limits and per-session caps. Do this before running any long automation.
- In your OpenRouter dashboard, click on the key you created.
- Set a hard limit — say $50 per month — and a soft limit that alerts you at $40.
- For Claude Code tasks that run on a schedule, add a
max_tokenscodecode parameter to your requests. A typical code generation task might use 4,000 output tokens. Set it explicitly so a runaway agent doesn’t burn through your budget.
Expected outcome: You stay within budget even if a script loops or a model hallucinates a massive response.
Step 7: Test with a Real Task
Run a small Claude Code job you already know the answer to — for example, refactor a Python function. Compare the output from direct API vs. OpenRouter. They should be identical in quality. Check the OpenRouter dashboard to confirm the token count matches.
Why this step is critical: The first run only proves it works once. The second run proves consistency. Run the same request twice and verify the output format matches. If OpenRouter adds any metadata or changes the response shape, adjust your parsing code.
The One Thing Most People Miss
OpenRouter’s real value isn’t the 3% price cut. It’s the ability to treat AI models as a pool rather than a single pipeline. When Anthropic Claude Code’s model is overloaded, you can dip into Gemini or Llama without changing a line of code. When a new model launches, you can test it by swapping one string in your environment variables.
The catch: OpenRouter adds a thin layer of abstraction. If Anthropic changes their API format or deprecates a version, OpenRouter has to update their routing. In practice, they keep pace, but it’s one more thing that can break. For production workloads, keep a direct key as a backup — not as your daily driver.
- Why are direct Claude Code API costs too high?Direct Anthropic API pricing locks you into a single provider, and costs can run away without spending caps.
- How does OpenRouter reduce Claude Code API costs?OpenRouter offers competitive pricing, automatic fallback, and the ability to route to cheaper models, reducing overall costs.
- How do you configure Claude Code to use OpenRouter?Set the ANTHROPIC_BASE_URL and ANTHROPIC_API_KEY environment variables to point to OpenRouter's endpoint.
