GET /api/arena/model-stats

​

Authentication

​

Response

curl -X GET 'http://localhost:5079/api/arena/model-stats' \
  -H 'Authorization: Bearer YOUR_JWT_TOKEN'

const response = await fetch('http://localhost:5079/api/arena/model-stats', {
  headers: {
    'Authorization': `Bearer ${jwt}`
  }
});

const stats = await response.json();

import requests

response = requests.get(
    'http://localhost:5079/api/arena/model-stats',
    headers={'Authorization': f'Bearer {jwt}'}
)

stats = response.json()

{
  "items": [
    {
      "model_id": "model-uuid-1",
      "model_name": "llama-3.3-70b-versatile",
      "display_name": "Llama 3.3 70B",
      "provider": "groq",
      "total_votes": 1523,
      "wins": 892,
      "losses": 431,
      "ties": 200,
      "win_rate": 0.585,
      "elo_rating": 1687,
      "last_voted_at": "2024-01-15T10:30:00.000Z"
    },
    {
      "model_id": "model-uuid-2",
      "model_name": "mixtral-8x7b-32768",
      "display_name": "Mixtral 8x7B",
      "provider": "groq",
      "total_votes": 1201,
      "wins": 645,
      "losses": 398,
      "ties": 158,
      "win_rate": 0.537,
      "elo_rating": 1542,
      "last_voted_at": "2024-01-15T09:15:00.000Z"
    }
  ]
}

{
  "error": "Unauthorized"
}

{
  "success": false,
  "error": "Database query failed",
  "code": "STATS_ERROR"
}

​

Side Effects

var stats = await _modelStatsService.GetModelStatsAsync();

• ai_models table: Model metadata
• model_votes table: Aggregated vote counts
• Potential JOIN with comparisons table (service-level)

• Aggregation query (COUNT, SUM operations assumed)
• Potentially expensive for large vote datasets

​

Authorization

• Any authenticated user

• Global statistics (not user-specific)
• All users see same data

​

Permissions

​

Response Structure

• Ordering algorithm not enforced by server contract
• No explicit sort parameter in controller

• Returns ALL models
• Could cause performance issues if many models

• No query parameters for filtering by provider, win_rate, etc.

​

Edge Cases

1. No votes exist: Behavior not enforced by server contract (empty array likely)
2. Model has zero votes: Inclusion behavior not enforced by server contract
3. Models without votes: Inclusion behavior not enforced by server contract
4. Inactive models: Inclusion behavior not enforced by server contract
5. Very large result set: No pagination, performance not guaranteed

​

Error Conditions

catch (Exception ex) {
    return StatusCode(500, new { error = ex.Message, code = "STATS_ERROR" });
}

• All service exceptions return 500
• Exception message exposed to client

​

Behavioral Guarantees

• Cache behavior implementation-defined and not guaranteed
• Real-time vs. cached not enforced by server contract

• Votes in progress may not be reflected
• Consistency model not enforced by server contract

​

Calculation Methods

• Calculation formula not enforced by server contract
• Behavior when total_votes = 0 not specified by API contract

• Calculation algorithm not documented in controller
• Null handling not enforced by server contract

• Vote counting logic not enforced by server contract
• Increment semantics implementation-defined and not guaranteed

​

Performance Characteristics

• Simple aggregation: Fast
• Complex Elo calculation: Slower

• model_votes.model_id should be indexed
• model_votes.created_at for last_voted_at

• Cache behavior implementation-defined and not guaranteed
• Cache invalidation strategy not specified by API contract

​

Use Cases

• Sort by win_rate or elo_rating
• Display top N models

• Used by GetTopperAndRandomModelAsync for dual-chat topper mode
• Selects highest-rated model

• Track model performance over time
• Compare provider effectiveness

​

Data Freshness

• Stats updated when vote recorded
• Update timing not enforced by server contract (immediate vs. batch)