Using your own API key for AI chatbots
Deskwoot lets you bring your own API key from OpenAI or Anthropic. This means you pay your provider's standard rates with no per-message fees from Deskwoot.
Updated September 30, 2026Also available in: Deutsch, Português
Bring your own OpenAI or Anthropic API key (BYOK) and pay zero to Deskwoot for AI usage. Available on Startup and above. Paste your key in Settings → Bots, pick the model, done — every AI conversation now bills directly to your account at the provider, bypassing our prepaid balance entirely. No quotas, no overage, no auto-recharge. The only limits are your own provider's rate limits.
Deskwoot lets you bring your own API key from OpenAI or Anthropic, giving you direct control over your AI chatbot costs. Instead of paying per-message fees to a third party, you pay your provider's standard rates,often resulting in significant savings. This approach is ideal for businesses that want transparency and predictable pricing as their support volume scales.
Supported providers
Deskwoot currently supports API keys from two leading AI providers:
- OpenAI: GPT-4o, GPT-4, and GPT-3.5 models
- Anthropic: Claude Sonnet and Claude Opus models
You can enter your API key directly in the bot settings within your Deskwoot dashboard. Once configured, your chatbot will use your account for all AI-powered responses.
How to set up your API key
- Obtain an API key from OpenAI or Anthropic by creating an account on their platform and navigating to the API section.
- Log in to your Deskwoot account and go to the bot settings page.
- Paste your API key into the designated field and select your preferred model.
- Save your settings. Your chatbot will now use your own API key for all conversations.
Cost comparison
When you use your own API key, typical costs range from $0.001 to $0.01 per message, depending on the model you choose and the complexity of the conversation. This is significantly cheaper than competitors who charge $0.99 to $2.00 per AI resolution. For high-volume support teams, this can translate to thousands of dollars in monthly savings.
Best practices
- Start with a cost-effective model like GPT-3.5 or Claude Sonnet to test performance before upgrading to more advanced options.
- Monitor your API usage through your provider's dashboard to track spending and set budget alerts.
- Review your chatbot's conversation logs regularly to ensure the AI is providing accurate and helpful responses.
Common troubleshooting
If your chatbot stops responding, verify that your API key is still valid and that your provider account has sufficient credits. API keys can expire or be revoked if billing information is outdated. Also check that you've selected a supported model in your bot settings.
Frequently asked questions
Can I switch providers after setting up my API key?
Yes, you can change your API key and provider at any time in the bot settings. Your chatbot will immediately start using the new configuration.
Will I be charged by both Deskwoot and my AI provider?
No. When you use your own API key, you only pay your AI provider's standard rates. Deskwoot does not add per-message fees on top of that.
What happens if my API key runs out of credits?
Your chatbot will stop responding until you add credits to your provider account. It's a good idea to set up billing alerts with your provider to avoid interruptions.
BYOK and AI conversation metering
When your live chat bot runs on your own OpenAI or Anthropic key, the conversations it serves do not count against the monthly AI conversations included with your plan. You pay your provider directly and Deskwoot does not meter those calls.
The Copilot tools an agent uses on the same conversation (suggest reply, summarize, translate, detect language) always run on Deskwoot-managed AI and count normally. If both the BYOK bot and Copilot touch the same conversation in the same month, that conversation counts once (because of the agent-side Copilot action) rather than twice.