Docs
Model Weights
Every AI model on OneConnecter, and the OCDs it costs to run. We don't mark up models — your weight reflects the model's cost, nothing more.
| Model | Provider | OCDs per call |
|---|---|---|
| Claude Haiku 4.5 | anthropic | 0.25 |
| GPT-4o-mini | openai | 0.25 |
| GPT-5-mini | openai | 0.25 |
| GPT-5-nano | openai | 0.25 |
| Gemini 2.5 Flash-Lite | 0.25 | |
| Gemini 3 Flash | 0.25 |
| Model | Provider | OCDs per call |
|---|---|---|
| Claude Sonnet 4.6 | anthropic | 0.5 |
| GPT-4o | openai | 0.5 |
| GPT-5 | openai | 0.5 |
| Gemini 3 Pro Preview | 0.5 |
| Model | Provider | OCDs per call |
|---|---|---|
| Claude Opus 4.7 | anthropic | 1 |
| GPT-5-pro | openai | 3 |
| OpenAI o3-pro | openai | 3 |
Most responses use 0.25–1 OCDs. Quick models use less, Premium models use more. You always see the cost in the model picker before you run anything.
Quick reply on Haiku
0.25 OCDs per response
Everyday reply on Sonnet
0.5 OCDs per response
Heavy reasoning on GPT-5-pro
3 OCDs per response
Weights are calculated from the model provider's actual per-call cost plus a small platform margin. When a model's cost goes down, its weight can go down with it. Weight changes only apply to new calls — your past usage is never recalculated.
Page refreshes hourly. New models appear here on or shortly after public availability — see our pricing page for plan details.