Chatwith supports multiple LLM models and APIs so you can create chatbots for different purposes and accomplish the best balance between speed, intelligence and cost.
Supported models
The tables below present all currently supported AI models on Chatwith. They have been selected by our team based on their overall high performance: speed, price, and reasoning capabilities.
A high number of black dots represents a comparatively high rating in each category.
Recommended models
Models picked by our team based on all-round, high performance-to-price ratio.
Model | Price | Speed | Reasoning & Actions | Context Size | Cost (message credits) |
OpenAI GPT-5-mini | ●●●●● | ●●●●● | ●●●●○ | 400k | 2 |
OpenAI GPT-5 | ●●○○○ | ●●●○○ | ●●●●● | 400k | 10 |
OpenAI GPT-4.1-mini | ●●●●● | ●●●●● | ●●●○○ | 1M | 2 |
OpenAI GPT-4.1 | ●●●○○ | ●●●●○ | ●●●●○ | 1M | 5 |
Anthropic Claude 4.5 Sonnet | ●●○○○ | ●●●○○ | ●●●●● | 1M | 10 |
Google Gemini Flash 2.5 | ●●●●○ | ●●●●● | ●●●○○ | 1M | 3 |
Mistral Medium | ●●●●○ | ●●●○○ | ●●●●○ | 128k | 3 |
Source of performance scores: https://artificialanalysis.ai
Deprecated models
Models which have been once available but are (or will be) deprecated.
Model | Cost (message credits) | Deprecation notes |
OpenAI GPT-4o-mini | 1 | Deprecated on November 1st 2025. |
OpenAI GPT-4o | 10 | Deprecated on November 1st 2025. |
Anthropic Claude 3.7 Sonnet | 10 | Deprecated on November 1st 2025. |
Google Gemini Flash 2.0 | 1 | Deprecated on November 1st 2025. |
Mistral Large | 5 | Deprecated on November 1st 2025. |
OpenAI GPT-3.5-Turbo | 1 | Deprecated on May 14th 2025. |
OpenAI GPT-4-Turbo | 10 | Deprecated on May 14th 2025. |
OpenAI GPT-4 | 20 | Deprecated on May 14th 2025. |
How to switch between models
You can change model used by your chatbot in by going to the Dashboard, then Settings and using the dropdown to choose a different AI model.
Changing models is possible only on our Hobby, Standard and Business plans.
How to use own OpenAI or OpenRouter API key
You can optionally use your own OpenAI or OpenRouter API key to bill your chatbot messages to. This helps with cost management.
If you provide your own API key, all messages will be sent using it (not our key). You’ll need to ensure that you have access to the chosen model and you’ll manage the API limits and billing yourself.
In other words: we are unable to guarantee 100% chatbot availability and performance if the provided key does not work (eg. is disabled by OpenAI).
