Cost Projection
1 creditGET /api/premium/cost/projectionThe cost projection endpoint takes your expected daily input + output token volume and projects the cost across up to 10 models. Returns daily/weekly/monthly/yearly totals per model plus a ranking by cheapest monthly. Pure compute on live pricing.
When to use this endpoint
When you need to budget for a workload across multiple model candidates. Useful for FinOps reporting or for an agent picking the cost-optimal model.
Parameters
| Name | In | Type | Description |
|---|---|---|---|
| model* | query | string | Comma-separated 1-10 model ids or names |
| input_tokens_per_day* | query | number | Expected daily input volume |
| output_tokens_per_day* | query | number | Expected daily output volume |
| horizon | query | string | Primary horizon: daily, weekly, monthly (default), yearly |
* required
Example response
{
"ok": true,
"workload": { "input_tokens_per_day": 1000000, "output_tokens_per_day": 500000 },
"projections": [
{
"model": "Claude Opus 4.7", "matched": true,
"rates": { "input_per_1m": 15, "output_per_1m": 75 },
"daily": { "total": 52.5 },
"monthly_total": 1575
}
],
"ranked_cheapest_monthly": [
{ "model": "Gemini 3", "monthly_total": 525 }
]
}Code samples
Python SDK
from tensorfeed import TensorFeed
tf = TensorFeed(token="tf_live_...")
p = tf.cost_projection(
models=["Claude Opus 4.7", "GPT-5.5"],
input_tokens_per_day=1_000_000,
output_tokens_per_day=500_000,
)TypeScript SDK
import { TensorFeed } from 'tensorfeed';
const tf = new TensorFeed({ token: 'tf_live_...' });
const p = await tf.costProjection({
models: ['Claude Opus 4.7'],
inputTokensPerDay: 1_000_000,
outputTokensPerDay: 500_000,
});MCP tool
Available via the TensorFeed MCP server as cost_projection. Add npx -y @tensorfeed/mcp-server to your Claude Desktop or Claude Code MCP config.