The Universal LLM Key
What is the Universal LLM Key?
The Universal LLM Key is a single API key that gives your projects access to 44+ language models across 7 providers, including current Claude, GPT, and Gemini model families, without requiring you to sign up for separate API accounts or manage individual keys for each provider.
Instead of juggling credentials from multiple AI providers, you use your Emergent Credits to fund model calls through the Universal Key. The platform handles routing, authentication, and billing automatically with no markup, you pay the platform's per-token rates from your credit balance.
Generate your key in Account Settings
The Universal LLM Key is managed in Account Settings → Universal API Key. You generate it yourself, it is not created automatically. Keys start with the prefix
. Access to advanced controls (rate-limit overrides, external proxy, etc.) requires a Standard or Pro plan; Free accounts are blocked from compute-intensive models.sk-emergent-
The key works inside your published apps, and on paid plans you can also use it as a LiteLLM-compatible proxy from your local machine or CI/CD pipelines.
Universal Key vs Your Own API Key
You have two options for connecting your projects to language models:
| Aspect | Universal Key | Your Own API Key |
|---|---|---|
| Setup | Generate once in Account Settings; no external accounts | Must create accounts with each AI provider |
| Billing | Charged in Emergent Credits; single balance | Billed directly by each provider |
| Model access | 44+ models / 7 providers via one key | Only models from the provider you signed up with |
| Rate limits | Shared platform limits | Provider's limits for your account tier |
| Cost transparency | Unified credit usage dashboard | Separate invoices from each provider |
| Markup | No markup on platform per-token rates | Provider list price |
When to use the Universal Key
- You're prototyping and want to try multiple models quickly
- You prefer consolidated billing and credit management
- You're building internal tools and don't want to manage external API keys
- You value simplicity over per-token cost optimization
When to bring your own key
- You've negotiated volume discounts with a specific provider
- You need model access beyond Emergent's supported set
- Your organization mandates direct contracts with AI providers
To bring your own LLM credentials, go to Account Settings → Universal API Key → Custom LLM Keys.
Switching is easy
You can start with the Universal Key and switch to your own API keys later by updating your project's Secrets. The agent will adapt automatically.
Recharging & Credit Types
The Universal Key draws from your Emergent Credit balance every time your app makes a model call. When your balance runs low, you'll need to recharge to keep services running.
How recharging works
Check your balance
View your current credit balance and usage history in Account Settings → Credit Usage.
Add credits
Purchase credits via your supported payment provider. Credits are added to your single combined balance instantly.
Credits are consumed
Each model call deducts credits based on token count and model pricing. You can see per-project breakdowns in the usage dashboard.
Credit types
- Subscription credits: Included with your paid plan each billing cycle; these do not roll over, they refill to the plan cap at each cycle.
- Purchased (top-up) credits: Pay-as-you-go; these never expire.
- Promotional/boost credits: Carry a shown fixed expiry date; applied before other credit types.
For full details on credit expiry and rollover rules, see Managing credit usage.
Auto-recharge available
Paid plans can enable auto-recharge to prevent service interruptions. When your balance drops below the configurable trigger threshold (minimum 5 credits), the platform automatically transfers funds from your Emergent credit balance to keep the Universal Key funded. Purchase-based auto-topup (charging your card automatically) is coming soon.
Using the Key Outside the Platform
On paid plans, you can use your Universal Key as a LiteLLM-compatible proxy endpoint to call models from your local development environment, scripts, or CI/CD pipelines.
The proxy base URL and full configuration details are shown in Account Settings → Universal API Key after you generate your key.
Setup example
Find your base URL
The exact proxy base URL is displayed in Account Settings → Universal API Key. Replace
<base_url_from_account_settings>Free tier limitation
The external proxy endpoint is not available on the Free tier, and Free accounts are blocked from compute-intensive models. Upgrade to Standard or Pro to use the Universal Key outside of published Emergent apps and to access advanced controls.
For security, the Universal Key proxy is IP-restricted. Requests must originate from known Emergent infrastructure or your verified NAT gateway. If you're on a dynamic IP, the platform may prompt you to verify your location the first time you call from a new network. Enterprise plans can whitelist specific CIDR ranges for CI/CD runners.
Regenerating the Key Breaks Live Apps
Your Universal Key is tied directly to your account and embedded in your published applications as the Secret
. If you regenerate it (for example, after a suspected leak), the old key is immediately invalidated.EMERGENT_LLM_KEY
Published apps will fail
Any live app that references the old Universal Key will start receiving 401 Unauthorized errors as soon as you regenerate. This includes:
- Apps making LLM calls via the Emergent SDK
- Integrations that use the key via the
environment variableEMERGENT_LLM_KEY - External scripts or services using the old key
Recovery steps
Regenerate the key
Go to Account Settings → Universal API Key and regenerate your key. Copy the new
sk-emergent-…Update Secrets
For each affected project, open Preview → Manage → Secrets, find the
EMERGENT_LLM_KEYRe-publish
Trigger a Re-publish so the updated Secret propagates to the live environment. Re-publishing is free of charge (beyond the monthly tier fee) .
Test in preview first
If your project has a preview environment, update the Secret there first and verify everything works before updating production. Note that preview and production environment variables are managed separately after the first publish.
When to regenerate
Only regenerate your Universal Key if:
- You believe the key has been exposed publicly (e.g., committed to a public repo)
- An ex-team member had access and you need to revoke it immediately
- You're responding to a security incident
For routine key rotation, coordinate the regeneration with a maintenance window to minimize downtime.
Supported Models
The Universal Key currently provides access to 44+ models across 7 providers. Current model families include:
- Claude: Fable 5.x / Opus 5 / Sonnet 5 / 4.6 / Haiku 4.5
- GPT: GPT-6 / GPT-5.x
- Gemini: 2.5-3.8
The full, up-to-date list of available models is shown in Account Settings → Universal API Key.
Free tier model restrictions
Free accounts are blocked from compute-intensive models. Upgrade to Standard or Pro to unlock the full model catalog.
