From request to response, in one unified flow
Every call you make travels through the FirmLink gateway — authenticated, routed, metered and returned — in a few hundred milliseconds. Here's the whole journey.
Create your API key
Access to FirmLink is invite-based to keep capacity predictable. Once registered, generate scoped, revocable API keys from the console in under a minute — each with its own usage view.
Point your SDK at FirmLink
Our API is fully compatible with the OpenAI SDK. Change the base URL and API key — your existing code, retries and streaming logic keep working untouched.
Call any model
A single request format reaches every model on the platform. Switch between GPT-5, Claude, Gemini or DeepSeek by changing one string — metering and failover happen automatically.
from openai import OpenAI
client = OpenAI(
base_url="https://api.firmlinktech.com/v1", # ← the only line that changes
api_key="YOUR_FIRMLINK_API_KEY",
)
for model in ["openai/gpt-6-astra", "anthropic/claude-sonnet-5", "deepseek/deepseek-v4.1-flash"]:
r = client.chat.completions.create(
model=model,
messages=[{"role": "user", "content": "Explain MaaS in one sentence."}],
)
print(model, "→", r.choices[0].message.content[:80])What happens to every request
Four stages, executed in milliseconds, on every single call — so you never have to build them yourself.
1. Auth
Every request is authenticated against your key's scopes and rate limits in under 5ms.
2. Rate limit
Token-bucket limiters protect your budget and keep noisy neighbors away from your capacity.
3. Smart router
Requests are matched to the healthiest, lowest-latency provider region for your model.
4. Metering
Tokens are counted on the way in and out, priced per model, and written to your ledger in real time.
Failover you never notice
Providers have bad days. Regions saturate. Deployments roll out and roll back. FirmLink continuously health-checks every provider route and shifts traffic away from degradation — typically within 300ms of the first failed probe.
Routing policy is yours to choose: optimize for latency, for price, or pin to a specific provider region for compliance. The default balanced policy suits most workloads.
Every routing decision is logged and visible in your console's activity feed, so there are never silent surprises in a postmortem.
One ledger for every model
Four providers, four invoices, four billing dashboards is the old way. On FirmLink, every token — regardless of model or provider — lands in a single itemized ledger, priced per model at transparent rates.
Top up once, draw down across everything. Export any month as CSV for your finance team.
Built for regulated workloads
TLS 1.3 everywhere
All traffic is encrypted in transit; keys are hashed at rest and shown only once at creation.
Zero retention
Prompts and completions pass through memory only. Nothing you send is stored or used for training.
Scoped keys
Issue per-service keys with independent limits, and revoke them instantly from the console.
Regional compliance
Operated from Dubai with UAE data-handling commitments and GDPR-aligned data rights.