fshy

Home / Cloudflare

Cloudflare Workers AI (Inference)

developers.cloudflare.com/workers-ai/platform/pricing/Inference endpoint, platform allocationinference/cloudflare-workers-aiJSONcomparereport an errorprovider JSON

Free allowance
10,000 Neurons per day at no charge on both the Free and the Paid Workers plan.
Verdict
caveatusable, with a named limit or an unprobed read path
Cost
free tier
Identity
emailno card, no phone
Binding limit
request-rate10,000 Neurons per day, reset at 00:00 UTC; further operations fail with an error past it; 300 requests per minute for Text Generation, a default per task type, unless the model requires the Workers Paid plan; models that require it...
Evidence
documented10 sources; a first-party page was read and the sentence quoted
Freshness
stale on 2026-11-29checked 2026-10-29 (live); stale on 2026-11-29
Free tier10,000 Neurons per day at no charge on both the Free and the Paid Workers plan. The page states the daily figure; the monthly figure of 300,000 and the token equivalents below are derived arithmetic on it, not vendor statements
unit tokens; metered on request-count; billing is in Neurons. The page prints a per-token price beside each model as an equivalent, so a single USD per million tokens figure for Workers AI does not exist. The free-plan input rate runs from 1,542 to 60,000 neurons per million tokens across 31 models, a factor of 38.9, so the token count behind one allowance is model-dependent: 194.6 million a month at the cheapest and 5.0 million at the dearest Derived figures and their bases: 300,000 = 10,000 x 30 days; 122.1 million input tokens = 300,000 / 2,457 x 1,000,000; 4.07 million a day = that divided by 30
Pricenone stated
Limits
  • request-rate (binding) 10,000 Neurons per day, reset at 00:00 UTC; further operations fail with an error past it source
  • request-rate (binding) 300 requests per minute for Text Generation, a default per task type, unless the model requires the Workers Paid plan; models that require it fall under the per-account, per-model limit in the next row. unverified: whether the 300 is counted per account is not stated for the default limits source
  • request-rate (binding) 20 requests a minute per account per model on standard billing for the eight paid-only models, 50 requests a minute with prepaid AI Gateway credits source
  • request-rate (binding) other task types, requests per minute: Automatic Speech Recognition 720; Image-to-Text 720; Text-to-Image 720 (@cf/runwayml/stable-diffusion-v1-5-img2img 1500); Translation 720; Image Classification 3000; Object Detection 3000; Summarization 1500; Text Classification 2000; Text Embeddings 3000 (@cf/baai/bge-large-en-v1.5 1500) source
Identityemail a Cloudflare account; the Workers Free plan is enough for the daily allocation
Egressnot published
Interfacerest
Toolingany OpenAI-compatible client; curl against POST {base}/chat/completions
Composes withpairs with a Cloudflare Worker already deployed; the 300,000-neuron monthly allowance is the scarce link, and it covers workload
Reliabilityeight named models require a paid billing method and fail on a Free plan
Notes300,000 Neurons a month, not 375 million: 10,000 a day times 30. At 2,457 neurons per million input tokens that is 122.1 million input tokens a month and 4.07 million a day. The output half alone is 2.3 times it. Every figure beyond the page's own 10,000 a day is arithmetic on it and is labelled as such in free_tier.stated.
Evidence notedocumented the allocation, the reset and the paid-only model list are quoted from the vendor pages read. No call was made, because that needs an account.
Parsed numbers
  • request-rate 300 requests / minute
  • request-rate 20 requests / minute
Evidence 10
datesourcewhat it proved
2026-10-22developers.cloudflare.com/workers-ai/platform/pricing/10,000 Neurons per day free, USD 0.011 per 1,000 Neurons above it on Workers Paid, all limits reset daily at 00:00 UTC, and eight named models require a paid billing method including @cf/moonshotai/kimi-k2.6 and @cf/zai-org/glm-5.3-flash
2026-10-22developers.cloudflare.com/workers-ai/platform/pricing/per-model neuron rates over 39 text-generation models, of which 31 are reachable on the free plan; input runs from 1,542 neurons per million tokens on @cf/ibm-granite/granite-4.0-h-micro to 60,000 on @cf/qwen/qwq-32b, and the eight paid-only ids run to 127,273 on @cf/zai-org/glm-5.3
2026-10-22developers.cloudflare.com/workers-ai/platform/limits/Text Generation is 300 requests per minute, 'unless the model requires the Workers Paid plan'; Image Classification, Object Detection and Text Embeddings are 3000; Summarization 1500; Text Classification 2000; and the eight paid models are 20 requests per minute on standard billing or 50 with prepaid AI Gateway credits
2026-10-28developers.cloudflare.com/workers-ai/platform/pricing/the free allocation is 10,000 Neurons per day, limits reset daily at 00:00 UTC, and operations past a limit fail with an error
2026-10-28developers.cloudflare.com/workers-ai/platform/limits/the paid-model limits are 20 requests per minute on standard billing and 50 with prepaid AI Gateway credits, per account and per model
2026-10-28developers.cloudflare.com/workers-ai/platform/pricing/the page names the models that require a paid billing method, kimi-k2.6 among them, and lists eight ids
2026-10-28developers.cloudflare.com/workers-ai/platform/limits/Text Generation is 300 requests per minute by default per task type, excluding models that require the Workers Paid plan
2026-10-28developers.cloudflare.com/workers-ai/platform/limits/the limits page lists per-task-type request rates, 720 for speech recognition, 3000 for image classification, 1500 for summarization and 2000 for text classification
2026-10-28developers.cloudflare.com/fundamentals/account/create-accouncreating a Cloudflare account asks for an email and a password, then sends an email to verify the address
2026-10-28developers.cloudflare.com/workers-ai/platform/pricing/the Workers Free plan carries the 10,000 Neurons per day free allocation
Unknown values 3
  • $.egress.metered egress metering was not published
  • $.per_object_cap.stated the field does not apply to this kind of offer
  • $.retention.stated the field does not apply to this kind of offer

Corrections