API Documentation

Authentication

All API requests require a session token. Get one by logging in:

POST /api/auth/login
Content-Type: application/json

{
  "email": "you@example.com",
  "password": "your-password"
}

→ Sets session cookie (httpOnly)
→ Or use: Authorization: Bearer {token}

Search

POST /api/search

Send a query to multiple AI models with web search. Returns NDJSON stream.

Request Body

Parameter	Type	Description
query required	string	Search query (max 500 chars)
models	string[]	Models: openai, gemini, perplexity, grok, claude (default: all)
mode	string	lite (Lite), standard (Standard), deep (Deep). Default: standard
moderatorModel	string	Moderator: grok (default), openai, gemini, perplexity, claude, deepseek, llama
images	array	Up to 3 images: [{mimeType, data(base64)}]

Credit Cost

Mode	Formula	Example (5 models)
Lite (minimal)	1 × model count	5 credits
Standard (basic)	models + 2	7 credits
Deep (advanced)	models × 2 + 3	13 credits

Response (NDJSON stream)

{"type":"answer","modelId":"openai","modelName":"ChatGPT","text":"..."}
{"type":"answer","modelId":"gemini","modelName":"Gemini","text":"..."}
{"type":"error","modelId":"grok","modelName":"Grok","message":"..."}
{"type":"confidence","score":4,"total":5,"label":"high"}
{"type":"summary","text":"...","moderatorName":"Grok"}
{"type":"complete","queryId":42}

Example (curl)

curl -X POST https://neuroverdict.com/api/search \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -d '{
    "query": "React vs Vue для SEO",
    "models": ["openai", "gemini", "perplexity"],
    "mode": "standard"
  }'

Query Classification

GET /api/search/classify?q={query}

Analyze query complexity and get suggested model count.

GET /api/search/classify?q=столица+Бразилии

→ {"complexity":"simple","suggestedModels":2,"reason":"factual"}

Balance & Plans

GET /api/payments/balance — Credit balance

GET /api/payments/plans — List available plans (public)

GET /api/payments/subscription — Current subscription

Query History

GET /api/history — List past searches

GET /api/history/:id — Get result with model responses

POST /api/history/:id/share — Generate public share link

GET /api/history/shared/:token — Get shared result (no auth)

Rate Limits

Endpoint	Limit
/api/search	20 requests/min
/api/auth/*	15 requests/15 min
/api/* (other)	60 requests/min

Error Codes

Code	Description
400	Invalid request (missing query, too long, invalid mode)
401	Not authenticated
402	Insufficient credits
429	Rate limit exceeded or concurrent search in progress
500	Server error