Works with every OpenAI-compatible provider
Drop your existing SDK. Point base_url at DalimServer. Done.
Built for real workloads
Auto rotation, live balances, model allowlists, per-key limits, playground, analytics — everything you need to run a production gateway.
Auto rotation
Exhausted key? Next one takes over instantly with zero downtime.
Multi-provider
Route to any OpenAI-compatible endpoint through a single master key.
Per-key limits
Set request and token caps per master key. Keep costs predictable.
Live analytics
Hourly charts of requests, tokens, cost, and error rate.
Playground
Test any model against any key from your browser. Instant feedback.
Model allowlist
Expose only the models you want — hide the other 460+.
User workspaces
Invite teammates. Give scoped access to specific providers and models.
Auto model
Let the gateway pick the cheapest/fastest model automatically.
Encrypted at rest
AES-256 encryption for every upstream key. Admin-approved sign-ups.
Two lines to switch
Change base_url and api_key. Every OpenAI SDK just works.
from openai import OpenAI
client = OpenAI(
base_url="https://dalimserver.lovable.app/v1",
api_key="YOUR_MASTER_KEY",
)
resp = client.chat.completions.create(
model="gpt-4o-mini",
messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)Questions, answered
Is my API key data safe?
All upstream keys are AES-256 encrypted at rest. Plaintext keys are only decrypted in-memory during proxy calls.
Which providers are supported?
Anything OpenAI-compatible — yunwu.ai, Groq, Cerebras, xAI, OpenRouter, DeepSeek, Gemini (compat mode), Anthropic (compat), Mistral, and any custom base URL you add.
How does auto rotation work?
On rate limit or exhausted balance, the gateway retries the same request with the next key in the pool — the client never sees the failure.
Do I need to sign up?
Yes. New accounts are queued for admin approval to keep the pool healthy.