Cloudflare Workers AI free allocation
Workers AI bills in neurons. The pricing page, last updated Oct 20, 2026, says the price is $0.011 per 1,000 Neurons. Anyone can use 10,000 Neurons per day at no charge. More than that requires the Workers Paid plan, which is then charged $0.011 per 1,000 Neurons above the same 10,000-Neuron daily allocation. Limits reset daily at 00:00 UTC. This allocation is not the Workers free-plan HTTP request quota.
| Plan | Free allocation | Price above it |
|---|---|---|
| Workers Free | 10,000 Neurons per day | N/A — upgrade to Workers Paid |
| Workers Paid | 10,000 Neurons per day | $0.011 / 1,000 Neurons |
Request limits by task
The limits page, last updated Sep 17, 2026, gives default per-task request rates. It does not say these RPM figures are free-plan-only. Local Wrangler inferences count toward the limits. Beta models may be lower.
| Task | Requests per minute |
|---|---|
| Automatic speech recognition | 720 |
| Image classification | 3000 |
| Image-to-text | 720 |
| Object detection | 3000 |
| Summarization | 1500 |
| Text classification | 2000 |
| Text embeddings | 3000; bge-large-en-v1.5 is 1500 |
| Text generation | 300, unless the model requires Workers Paid |
| Text-to-image | 720; stable-diffusion-v1-5-img2img is 1500 |
| Translation | 720 |
Models that require the Workers Paid plan do not get that default. Per account, per model, standard Workers AI billing is 20 requests per minute, and prepaid AI Gateway credits are 50 requests per minute.
Models the free allocation cannot buy
The pricing page says these models require a paid billing method, either Workers Paid or prepaid AI Gateway credits: @cf/moonshotai/kimi-k2.6, @cf/moonshotai/kimi-k2.7-code, @cf/zai-org/glm-5.2, @cf/zai-org/glm-5.3, @cf/zai-org/glm-5.3-flash, @cf/deepseek-ai/deepseek-v4-flash-0731, @cf/deepseek-ai/deepseek-v4.1-flash-1017, and @cf/deepseek-ai/deepseek-v4-pro-0813. Workers Free cannot purchase neurons above 10,000 per day.
Data use
The data-usage page does not split free and paid. It says Workers AI processes Customer Content only as long as needed to provide the service, and for standard inference that does not invoke a stateful feature Cloudflare does not retain inputs or outputs after the response is delivered, subject to the exceptions on that page. Cloudflare does not use Customer Content processed by Workers AI to train AI models. Exceptions include storage you select (R2, KV, Durable Objects, Vectorize), AI Gateway logging or caching, and async, batch, file, or adapter features. Operational metadata such as token or neuron counts may be retained and does not include Customer Content. Third-party models may have their own license terms.
How to get a token
The REST API guide says to sign up for Workers, open Workers AI, choose Use REST API, create a Workers AI API token, and copy the account id. A custom token needs Workers AI Read and Workers AI Edit. Requests go to https://api.cloudflare.com/client/v4/accounts/{account_id}/ai/run/{model} with Authorization: Bearer.
Other free-tier notes
Sources
Last checked 2026-10-22.