One API key for all AI models

Most AI platforms force separate keys per model or per provider. unoblox issues one key (ub-gw-…) that unlocks 50+ models at any time. Unified billing, shared quotas, one endpoint.

Why One Key Matters

• Simplicity: No env-var sprawl (UB_GPT_KEY, UB_CLAUDE_KEY, …)
• Cost control: Pool all model spend under one budget; see the true AI cost
• Rate limiting: Share quotas intelligently (e.g., 10k requests/day split across all models)
• Audit trail: Every model call logged against one key for compliance + cost recovery
• Billing: One ₹ invoice for Qwen, GPT, Claude, Llama combined

Your Dashboard

Component	What you see
Models	50+ listed; select what you use
Budget	₹ per month, tracked across all models
Usage	Requests/tokens per model + total
API keys	One production key, optional test key
Logs	Export for 90–180 days per key

No per-model API key rotation. Generate a key once; it works forever (or until you rotate for security).

How It Works in Code

# One key, many models
client = OpenAI(api_key="ub-gw-12345...", base_url="https://api.unoblox.ai/v1")

# Call any model in your catalog
response = client.chat.completions.create(
  model="qwen/qwen3-1.7b",      # FREE
  messages=[{"role": "user", "content": "..."}]
)

# Switch to a paid model—same key
response = client.chat.completions.create(
  model="openai/gpt-4o",          # ₹252/1M input
  messages=[...]
)

# Later, add Claude—same key still works
response = client.chat.completions.create(
  model="anthropic/claude-opus",   # ₹504/1M input
  messages=[...]
)

All calls bill against the same key, appear on one monthly invoice in rupees.

Key Rotation & Security

Rotate keys anytime via the dashboard. Old key revokes instantly; new key inherits the same model permissions. Rate-limit per key, not per model. Suitable for multi-tenant apps (one key per customer, separate budgets).

Frequently asked questions

Can I use the same key across multiple apps?

Yes, but not recommended. Each app should have its own key for audit isolation. Rotation/revocation becomes messy across codebases.

Do I need separate keys for dev/staging/prod?

Optional. Use one key everywhere, rely on model selection + code branch logic. Or generate separate keys for isolation + independent rate limits.

What if my key is compromised?

Revoke instantly in the dashboard. Spending stops immediately. Old key expires; new key is auto-assigned. Check usage logs for unauthorized calls since last check.

Can I limit which models one key can access?

Not yet. unoblox gives the same key access to all enabled models. Per-key model restrictions are on the roadmap.

Do you charge per key, or per-model?

Per model, per token. One key is free. Billing is by inference volume (₹ per million tokens, input/output), regardless of how many keys or models you layer on top.

How do I know my key has the right permissions?

Try calling each model once in a test script. unoblox returns 401 Unauthorized if the key lacks access (rare). Otherwise, you're good.

Get started in rupees → https://unoblox.ai/sign-in