OpenAI- and Anthropic-compatible endpoints. Programmatic key management. Per-key budgets, model restrictions, and region controls. All the compliance automation your production app needs — without building it yourself.
Need Codex-native behavior for Codex ACP or other Codex-specific clients? AI Mission Control also exposes a native Codex endpoint at https://taas.cloudsigma.com/v1/codex, separate from the standard OpenAI-compatible base URL.
Use /v1 for OpenAI-compatible SDK traffic. Use /v1/codex for Codex-native clients that expect native request semantics. Anthropic SDK clients use POST /v1/messages on the same base URL — change only the base URL and API key.
Drop-in OpenAI SDK replacement
from openai import OpenAI
# One line change — everything else works
client = OpenAI(
api_key="your-taas-key",
base_url="https://taas.cloudsigma.com/v1"
)
response = client.chat.completions.create(
model="gpt-4o",
messages=[{
"role": "user",
"content": "Hello!"
}]
)
print(response.choices[0].message.content)Compatible with OpenAI and Anthropic SDKs — Python, Node.js and Go
Use managed authentication, budgets and reliable model access through AI Mission Control.
Issue scoped API keys via the Management API. Integrate key provisioning into your onboarding flow, CI/CD pipeline, or infrastructure-as-code — no manual console steps needed.
Set a hard budget limit and period (daily, monthly, rolling) on every key. When the cap is hit, the key stops — protecting you from runaway costs or compromised credentials.
Lock each key to an explicit list of models. A key issued for a fast summarisation workload cannot call a more expensive reasoning model — enforced server-side.
Restrict keys to EU or other regions. Add "region": "EU" per request for call-level enforcement. AI Mission Control hard-fails requests that cannot be routed compliantly — no silent data leakage.
CloudSigma manages reliability for the selected model within your access policies. Applications receive a compatible response or a documented error they can handle.
Full structured audit logs per key — model, tokens, cost, region, supplier, latency. Export to your SIEM or pull via API for compliance evidence without building your own logging pipeline.
AI Mission Control implements the full OpenAI REST API. Every official SDK — Python, Node.js, Go, Ruby, Java — works by changing a single line: the base URL.
All endpoints are supported: /chat/completions, /embeddings, /audio/speech, /audio/transcriptions, /rerank and more. Streaming, function calling, and JSON mode work exactly as expected. The Anthropic Messages API (/v1/messages) runs in parallel for Anthropic SDK clients.
// OpenAI-compatible SDKs
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.TAAS_KEY,
baseURL: "https://taas.cloudsigma.com/v1",
});
// Codex-native clients (for example Codex ACP)
// should target the dedicated Codex base URL instead:
// https://taas.cloudsigma.com/v1/codex
// Streaming works too
const stream = await client.chat.completions.create({
model: "claude-sonnet-4-5",
stream: true,
messages: [{ role: "user", content: "Tell me a story." }],
});
for await (const chunk of stream) {
process.stdout.write(
chunk.choices[0]?.delta?.content ?? ""
);
}Issue scoped keys for users, tenants, or workloads from your own backend. Each key carries its own budget, model list, region, and rate limit — no manual steps in the AI Mission Control console required.
POST /v1/manage/keys
{
"name": "tenant-acme-prod",
"allowed_models": [
"gpt-4o-mini",
"text-embedding-3-small"
],
"allowed_regions": "EU",
"budget_limit": 100.00,
"budget_period": "monthly",
"rate_limit_rpm": 60,
"metadata": {
"tenant_id": "acme",
"env": "production"
}
}
// Response
{
"id": "key_abc123xyz",
"key": "taas-sk-...",
"name": "tenant-acme-prod",
"created": "2024-01-15T10:00:00Z"
}Query current usage and remaining budget for any key at any time. Build spend dashboards, trigger alerts, or raise limits programmatically — all via API.
GET /v1/manage/keys/key_abc123xyz
// Response
{
"id": "key_abc123xyz",
"name": "tenant-acme-prod",
"budget_limit": 100.00,
"budget_used": 43.17,
"budget_remaining": 56.83,
"budget_period": "monthly",
"resets_at": "2024-02-01T00:00:00Z",
"status": "active"
}
// Update budget limit inline
PATCH /v1/manage/keys/key_abc123xyz
{ "budget_limit": 200.00 }For GDPR-sensitive workloads, add "region": "EU" to every call — or lock the key to allowed_regions: "EU" so no request can ever leave European infrastructure, even if the calling code forgets the field.
Full structured audit logs — model, tokens, cost, supplier, region, latency — are available per key via the Management API, giving you the evidence trail your DPO needs without building a separate logging pipeline.
# EU-only embedding for GDPR compliance
curl https://taas.cloudsigma.com/v1/embeddings \
-H "Authorization: Bearer $TAAS_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "text-embedding-3-large",
"region": "EU",
"input": "Patient record summary..."
}'
# Audit log entry returned via Management API
{
"timestamp": "2024-01-15T14:32:01Z",
"key_id": "key_abc123xyz",
"model": "text-embedding-3-large",
"supplier": "openai-eu",
"region": "EU",
"input_tokens": 78,
"output_tokens": 0,
"cost_usd": 0.00031,
"latency_ms": 221
}Use the OpenAI or Anthropic model ID from the catalogue with your CloudSigma API key. CloudSigma manages service reliability while your application handles documented terminal errors.
# Python — use your CloudSigma key and the catalogue model ID
from openai import OpenAI
client = OpenAI(
api_key="your-taas-key",
base_url="https://taas.cloudsigma.com/v1"
)
response = client.chat.completions.create(
model="gpt-6-astra",
messages=[{"role":"user","content":"Analyse this."}]
)
print(response.choices[0].message.content)
# Handle terminal errors using the documented status and error code.No new SDKs to learn. No credential juggling. One API for everything.
Create a AI Mission Control account and retrieve your root admin key. Store it in your secrets manager — this key can create and revoke all other keys.
Use POST /v1/manage/keys to create a key per environment or tenant — with the exact model list, region, budget, and rate limit each workload needs.
Change base_url to https://taas.cloudsigma.com/v1 and swap in your scoped key. Every existing API call works immediately — no other changes required.
Pull real-time usage and audit logs via API. Adjust budgets or rate limits without reissuing keys. Revoke instantly when a key is no longer needed.
One endpoint for your models. AI Mission Control manages model access, budgets and audit logs so you can focus on building features.