Console

Foundry console at https://platform.sagea.space is the self-serve home for SAGEA Studio. Create a workspace, issue keys, try models without code, and watch spend.

In this guide

• Create a workspace and invite a teammate
• Generate and rotate API keys for development
• Compare models in the Playground without code
• Read usage dashboards and set spend alerts

Create a workspace

Sign in at https://platform.sagea.space with your work email and create a workspace for your team or project. One workspace per product environment keeps usage and keys separate, for example acme-dev and acme-prod.

Name the workspace after the app, pick the Kathmandu region default unless latency testing says otherwise, and invite one teammate as backup so keys are never locked to a single inbox.

# Workspaces isolate keys and usage per environment
# acme-dev  -> prototyping with sage-2-5-celer
# acme-prod -> live traffic with sage-2-4-actus
echo "Use acme-dev key locally, acme-prod key in deployment"

Workspace settings also show the base URL https://api.sagea.space/v1 used by every Studio example. New to the API flow? Start with First API request.

Generate and rotate API keys

Create a key per environment and store it in $SAGEA_API_KEY. Never paste keys into chat logs, screenshots, or client-side code. If a key leaks, revoke it in the console and issue a replacement.

# Use the key from the console
export SAGEA_API_KEY="sk-live-..."
 
curl -X POST https://api.sagea.space/v1/chat/completions \
  -H "Authorization: Bearer $SAGEA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "sage-2-5-celer", "messages": [{"role": "user", "content": "नमस्ते!"}]}'

Rotate on a schedule your team will keep, such as every 90 days and on every departure:

1. Generate a new key in console under Developers, Keys, Create key.
2. Deploy the app with both keys accepted for one release.
3. Switch traffic to the new key and confirm error rates hold.
4. Revoke the old key and delete it from vaults and .env files.

Keep dev, staging, and prod keys distinct so a local script can never bill production. See Platform overview.

Try the Playground without code

The Playground compares sage-2-4-actus versus sage-2-5-celer side by side on the same prompt. Paste a Nepali customer question once, run both models, and judge tone, length, and grounding before writing code.

A useful first test is a grounded invoice question with temperature low. Check which model cites Rs. 24,500 correctly and which adds unsupported detail. Copy the winning settings into your app.

# Reproduce the Playground winner in code
import os
import requests
 
resp = requests.post(
    "https://api.sagea.space/v1/chat/completions",
    headers={
        "Authorization": f"Bearer {os.environ['SAGEA_API_KEY']}",
        "Content-Type": "application/json",
    },
    json={
        "model": "sage-2-4-actus",  # winner from Playground
        "temperature": 0.2,
        "messages": [
            {"role": "system", "content": "Answer only from the provided context."},
            {"role": "user", "content": "Invoice total Rs. 24,500. Question: कुल रकम कति हो?"},
        ],
    },
    timeout=60,
).json()
print(resp["choices"][0]["message"]["content"])

Need help choosing? Read Choosing a model, then compare costs in Choosing a Model.

Read usage and set spend alerts

The Usage page graphs requests, tokens, audio minutes, and OCR pages per workspace. Filter by day and model to spot a runaway script, for example a loop calling sonus-tts with the same prompt.

Set a monthly spend alert at 50 percent and 100 percent of budget so the team gets email before the card does. Start with a low cap on acme-dev, review weekly, then raise the cap only after launch traffic looks clean.

For endpoint shapes behind the graphs, see Chat Completions, Text to Speech, and OCR Process.

Best practices

• Keep one workspace per environment with separate API keys.
• Store keys only in $SAGEA_API_KEY and secret managers, never in repos.
• Rotate keys every 90 days and immediately after any leak.
• Validate sage-2-4-actus versus sage-2-5-celer in the Playground first.
• Set spend alerts early and review usage dashboards weekly.

What's next

• First API request
• Choosing a model
• Foundry

Need centralized controls or procurement? See Foundry or contact [email protected].