De.Ciphyr evaluates models weekly on architecture-specific tasks — schema design, feature decomposition, and tech spec quality. Pricing adjusts automatically so you always get fair value.
1 credit = $0.01 USD. When you make an API request, the cost is calculated from actual token usage at the model's current rate.
Premium models (higher benchmark scores) cost more credits per generation but produce higher-quality output. Budget models cost less but may require more iterations.
You only pay for what you use. If a generation costs less than the estimate, the difference is refunded to your wallet automatically.
Sorted by quality score
| # | Model | Provider | Benchmark | Tokens/Credit | Input $/M | Output $/M | Tier |
|---|---|---|---|---|---|---|---|
| #1 | Claude Sonnet 4.6 claude-sonnet-4.6 | anthropic | 91/100 | 91 | $3.00 | $15.00 | Premium |
| #2 | Gemini 3.1 Pro gemini-3.1-pro | 87/100 | 109 | $2.00 | $12.00 | Premium | |
| #3 | GPT-4o gpt-4o | openai | 84/100 | 132 | $2.50 | $10.00 | Premium |
| #4 | Gemini 2.5 Flash gemini-2.5-flash | 72/100 | 555 | $0.30 | $2.50 | Standard | |
| #5 | Claude Haiku 4.5 claude-haiku-4.5 | anthropic | 68/100 | 270 | $1.00 | $5.00 | Standard |
| #6 | GPT-4.1 Nano gpt-4.1-nano | openai | 62/100 | 3,333 | $0.10 | $0.40 | Budget |
Benchmark scores are evaluated weekly on De.Ciphyr-specific tasks: schema design accuracy, feature decomposition quality, and tech spec completeness. Rates adjust automatically.
Estimated cost for common API use cases
Estimates are based on average token usage. Actual costs may vary depending on prompt complexity. Overpayments are automatically refunded to your wallet after each generation.
Free tier includes 15 credits. Pro unlocks all models, MCP integration, and full API access.