AI Gateway - FLIN Docs

One Line to AI

No SDKs. No configuration. Just call:

                        
ai-example.flin

                    // Claude (Anthropic)
answer = ask_claude("What is FLIN?")

// OpenAI
answer = ask_openai("Write a haiku about programming")

// Grok (X.AI)
answer = ask_grok("Tell me a joke")

// Mistral
answer = ask_mistral("Translate to French: Hello")

// FLIN's own AI (api.flin.sh)
answer = ask_flin("Generate a User entity")

Automatic Fallback

Use ask_ai() to automatically try providers in order:

                        
fallback.flin

                    // Tries: Claude → OpenAI → Grok → Mistral → FLIN
answer = ask_ai("Explain quantum computing simply")

print(answer)

💡 Smart Fallback

If Claude is down or rate-limited, FLIN automatically tries the next available provider. No code changes needed.

With Options

Pass an options object for more control:

                        
with-options.flin

                    answer = ask_claude("Write production-ready code", {
    model: "claude-sonnet-4-20250514",
    max_tokens: 4000,
    temperature: 0.3,
    system_prompt: "You are a senior Rust developer."
})

Available Options

Option	Type	Default	Description
`model`	text	Provider default	Model to use
`max_tokens`	int	1024	Maximum response tokens
`temperature`	float	1.0	Creativity (0.0 - 2.0)
`system_prompt`	text	none	System instructions

Supported Providers

Function	Provider	Default Model	Env Variable
`ask_claude()`	Anthropic	claude-sonnet-4-20250514	`ANTHROPIC_API_KEY`
`ask_openai()`	OpenAI	gpt-4o	`OPENAI_API_KEY`
`ask_grok()`	X.AI	grok-3	`XAI_API_KEY`
`ask_mistral()`	Mistral	mistral-large-latest	`MISTRAL_API_KEY`
`ask_flin()`	FLIN Cloud	flin-coder	`FLIN_API_KEY`

API Keys Configuration

Set your API keys as environment variables:

                        
~/.zshrc or ~/.bashrc

                    # AI Gateway Keys
export ANTHROPIC_API_KEY="sk-ant-api03-..."
export OPENAI_API_KEY="sk-proj-..."
export XAI_API_KEY="xai-..."
export MISTRAL_API_KEY="..."

# FLIN Cloud
export FLIN_API_KEY="flin_..."

⚠ Security

Never commit API keys to git. Use .env files (add to .gitignore) or environment variables.

Built-in Features

Caching

LRU cache with 1-hour TTL. Same prompt = instant response. No extra API costs.

Retry Logic

Exponential backoff with jitter. Handles 429 rate limits and 503 errors automatically.

Circuit Breaker

Per-provider circuit breaker. If a provider fails 5 times, it's bypassed for 60 seconds.

Validation

Prompt size limits, token bounds, temperature range (0.0-2.0). Prevents wasted API calls.

Real Example: AI Chat

                        
                        
                        
                    
app/index.flin

                    question = ""
answer = none
loading = false

fn askQuestion() {
    if question.len > 0 {
        loading = true
        answer = ask_ai(question)
        loading = false
    }
}

<div class="chat">
    <input
        placeholder="Ask anything..."
        value={question}
        enter={askQuestion()} />

    <button click={askQuestion()} disabled={loading}>
        {if loading then "Thinking..." else "Ask"}
    </button>

    {if answer}
        <div class="response">{answer}</div>
    {/if}
</div>
                

Where to Get API Keys

Provider	Dashboard
Anthropic (Claude)	console.anthropic.com
OpenAI	platform.openai.com
X.AI (Grok)	console.x.ai
Mistral	console.mistral.ai
FLIN Cloud	flin.dev/api-keys