Cloudflare Workers AI
Run frontier open-weight models on Cloudflare's global GPU fleet.
Cloudflare Workers AI runs open-weight models on Cloudflare's edge infrastructure. It requires two credentials: a account ID and an API token with Workers AI permissions. Scira's /model picker filters to models that declare function_calling: true in the Cloudflare catalog — required for the agent loop to work.
Get credentials
Find your account ID
Sign in to dash.cloudflare.com. On the Workers & Pages overview page, your Account ID is in the right-hand sidebar — a 32-character hex string.
Create an API token
Go to My Profile → API Tokens → Create Token. Use the Edit Cloudflare Workers AI template (or a custom token with Workers AI: Read permission). Click Create Token and copy it.
Add both to Scira
scira init
# or manually:
echo "CLOUDFLARE_ACCOUNT_ID=your-account-id" >> ~/.scira/.env
echo "CLOUDFLARE_API_TOKEN=your-token" >> ~/.scira/.envConfigure
{
"llmProvider": "workers-ai",
"model": "@cf/moonshotai/kimi-k2.6"
}Switch to Workers AI from the TUI with /llm → Cloudflare Workers AI.
Available models
The /model picker fetches the live model list and filters to function-calling capable models. Current confirmed models with tool use:
| Model ID | Description |
|---|---|
@cf/moonshotai/kimi-k2.6 | Default. 1T-parameter frontier model, 262k context, multi-turn tool calling |
@cf/zai-org/glm-4.7-flash | Fast multilingual model, 131k context, multi-turn tool calling |
@cf/ibm-granite/granite-4.0-h-micro | IBM Granite 4.0, strong on instruction following and agentic tasks |
@cf/qwen/qwq-32b | Qwen reasoning model, 32B parameters |
@cf/openai/gpt-oss-120b | OpenAI open-weight model, 120B, reasoning and agentic tasks |
The full catalog with capability tags is at developers.cloudflare.com/workers-ai/models. Filter by Function Calling to see the current full list.
Pricing
Workers AI has a free tier (10,000 neurons/day). See developers.cloudflare.com/workers-ai/platform/pricing for current limits and paid plan rates.