The Shape API

All requests require a bearer token. Get one from the API keys page.

Authorization: Bearer sk-shape-...

Keys are shown once at creation time. We only store a hash; if you lose a key, revoke it and create a new one.

The base URL is:

https://jointheshape.com/api/v1

A minimal call looks like this:

curl https://jointheshape.com/api/v1/models/peptide-1/predict \
  -H "Authorization: Bearer $SHAPE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"sequence":"MKWVTFISLLFLFSSAYS","task":"predict_binding","target":"GLP1R"}'

In Python:

import os, requests

resp = requests.post(
    "https://jointheshape.com/api/v1/models/peptide-1/predict",
    headers={"Authorization": f"Bearer {os.environ['SHAPE_API_KEY']}"},
    json={"sequence": "MKWVTFISLLFLFSSAYS", "task": "predict_binding", "target": "GLP1R"},
    timeout=60,
)
print(resp.json())

The API returns HTTP status codes and a JSON error envelope:

{ "error": { "type": "insufficient_credits", "message": "..." } }

• 401 authentication_error — missing or invalid bearer token
• 402 insufficient_credits — top up at /dashboard/billing
• 404 model_not_found — wrong model id
• 502 model_error — upstream model failure (credits refunded)
• 503 model_unavailable — model not currently serving

Each call is billed at a fixed per-call rate, deducted from your prepaid credit balance. Failed model calls are automatically refunded.

See the pricing page for current rates.

De novo peptide design & property prediction

Peptide-1 generates and evaluates peptide sequences for binding affinity, stability, and therapeutic potential. Trained on curated protein-peptide interaction data.

Endpoint

POST /api/v1/models/peptide-1/predict

Request body

{
  "sequence": "MKWVTFISLLFLFSSAYS",
  "task": "predict_binding",
  "target": "GLP1R"
}

Response

{
  "id": "sync-…",
  "model": "peptide-1",
  "output": {
    "binding_affinity_kcal_mol": -8.7,
    "stability_score": 0.82,
    "properties": {
      "hydrophobicity": 0.41,
      "charge": -1
    }
  },
  "usage": {
    "cost_usd": 0.01,
    "latency_ms": 1240
  }
}