Compare models for your workload.
See current token rates, check the features your application needs, and choose a model using your own tasks.
Current OpusGate rates
| Model | Input | Output | Cache read | Cache write |
|---|---|---|---|---|
| Claude Sonnet 5.5 | $0.78 | $3.90 | $0.09 | $0.96 |
| GPT-6.2 Sol | $0.42 | $2.16 | — | — |
| Gemini 3.8 Flash | $0.30 | $1.44 | $0.03 | $0.30 |
| Claude Opus 5.5 | $1.29 | $6.48 | $0.15 | $1.00 |
| DeepSeek V4 Pro | $0.24 | $0.68 | $0.24 | $0.24 |
| Qwen 3.8 Flash | $0.12 | $0.36 | $0.12 | $0.12 |
| AWS Claude Haiku 5.5 | $0.27 | $0.42 | — | — |
| AWS Claude Opus 5.5 | $6.00 | $9.60 | — | — |
OpusGate rates in USD per 1M tokens. A dash means no separate rate is published. Check model details for supported capabilities; input, output and cache rates can differ.
What to compare
Cost per complete request. Input and output are priced separately. Long responses, repeated context and available caching can change the total.
Required capabilities. Check the model details for supported vision, thinking and streaming features. Missing information is shown as unknown rather than assumed support.
Results on your tasks. Use a small, representative set of prompts to compare quality and latency. Keep the same input and output budget when comparing spend.
Integration. An API-compatible endpoint does not guarantee every model supports every client feature. Start with the API reference and your tool’s setup guide.
The full catalog includes model details and the current availability of image and video generation.
Questions about pricing
Can I use different models with one API key?
Yes. Select an available model ID in your request. Supported parameters and API protocols depend on the selected model.
Which model costs least?
Compare input, output and any published cache rates for your expected workload. A lower input rate alone does not determine the cost of a complete request.
Which model should I choose for coding?
Check that the model supports the features your tool needs, then test it on a representative task. Compare response quality, tool use, latency and actual billed spend.
Are these official provider prices?
No. The table shows OpusGate retail rates. It does not estimate or multiply an official provider price.
Connect the model you choose
Create a key, configure your client, and use your prepaid balance for requests.