Cloudflare Workers AI
Cloudflare Workers AI runs AI models at the edge, on Cloudflare's global network of data centers. This means inference happens at the server closest to you, minimizing latency. With a generous free tier of 10,000 neurons per day (no credit card required), it is one of the easiest ways to start using AI models.
Getting an API Key
- Visit dash.cloudflare.com/profile/api-tokens
- Create a new API token with Workers AI read permissions
- Copy your Account ID from the Cloudflare dashboard sidebar
- Paste both the API token and Account ID into AISCouncil under Settings > AI Model > Cloudflare
Cloudflare Workers AI offers 10,000 neurons per day for free with no credit card required. This is enough for casual usage across multiple models. Paid usage beyond the free tier is billed per neuron.
API keys are stored locally in your browser (localStorage) and are never included in shared bot URLs.
Supported Models
| Model | Context | Max Output | Price | Capabilities |
|---|---|---|---|---|
| Llama 3.3 70B | 128K | 4K | Free | Tools, code, streaming |
| Llama 4 Scout | 128K | 4K | Free | Vision, tools, code, streaming |
| Qwen 2.5 Coder 32B | 32K | 4K | Free | Tools, code, streaming |
| QwQ 32B | 32K | 4K | Free | Reasoning, streaming |
| DeepSeek R1 Distill 32B | 32K | 4K | Free | Reasoning, code, streaming |
| Gemma 3 12B | 32K | 4K | Free | Vision, streaming |
| Llama 3.1 8B | 8K | 2K | Free | Tools, code, streaming |
| Mistral 7B | 32K | 2K | Free | Streaming |
All models are free within the 10K neurons/day limit. See developers.cloudflare.com/workers-ai/platform/pricing for paid tier rates.
- Llama 3.3 70B is the largest and most capable free model -- great for general-purpose tasks.
- QwQ 32B and DeepSeek R1 Distill 32B offer free reasoning capabilities.
- Llama 4 Scout and Gemma 3 12B support vision input for free.
- Qwen 2.5 Coder 32B is a strong free code specialist.
Edge Inference
Cloudflare runs models on its global network of 300+ data centers. Your request is routed to the nearest location with GPU capacity, which means:
- Low latency -- inference happens close to you geographically
- No cold starts -- models are kept warm across the network
- Global availability -- works well from any region
Account ID Requirement
Unlike other providers that only need an API key, Cloudflare Workers AI also requires your Account ID. This is a 32-character hex string found in your Cloudflare dashboard sidebar. The base URL includes your account ID:
https://api.cloudflare.com/client/v4/accounts/{account_id}/ai/v1/chat/completions
AISCouncil prompts you for both the API token and Account ID during setup.
Configuration
Select Cloudflare as the provider when creating a bot profile. Enter both your API token and Account ID. The app constructs the correct endpoint URL automatically.
Tips for Best Results
- Start with Cloudflare for zero-cost experimentation. The free tier with no credit card requirement makes it the lowest-friction way to try AI models.
- Use reasoning models (QwQ, DeepSeek R1 Distill) for free chain-of-thought. These are some of the only free reasoning models available from any provider.
- Be aware of the neuron budget. Complex models and long prompts consume more neurons. Monitor your daily usage in the Cloudflare dashboard.