Cloudflare Workers AI
Run AI calls close to Workers when the backend is already on Cloudflare.
- Requests / min
- 300
- Neurons / day
- 10K
- Card required
- No
Cloudflare Workers AI scores 7.8/10 for free value in LLM. The main constraint: Free allocation is 10,000 neurons/day, resetting daily at 00:00 UTC; text generation is separately rate-limited at 300 requests/minute.
Free value score
7.8 / 10Show the math
Key metrics
Models
40 free · 68 totalLimits may be stale — last re-verified outside this section's freshness window.
| Model | Family | AA | Access |
|---|---|---|---|
| @cf/moonshotai/kimi-k2.6 | Kimi | 42.8 | free |
| @cf/moonshotai/kimi-k2.5 | Kimi | 38.1 | free |
| @cf/zai-org/glm-4.7-flash | GLM | 33.8 | free |
| @cf/google/gemma-4-26b-a4b-it | Gemma | 25.7 | free |
| @cf/openai/gpt-oss-120b | GPT | 23.8 | free |
| @cf/openai/gpt-oss-20b | GPT | 14.9 | free |
| @cf/deepseek-ai/deepseek-r1-distill-qwen-32b | DeepSeek | 11 | free |
| @cf/meta/llama-4-scout-17b-16e-instruct | Llama | 10 | free |
| @cf/qwen/qwen2.5-coder-32b-instruct | Qwen | 7.1 | free |
| @cf/aisingapore/gemma-sea-lion-v4-27b-it | Gemma | — | free |
| @cf/google/gemma-2b-it-lora | Gemma | — | free |
| @cf/google/gemma-3-12b-it | Gemma | — | free |
| @cf/google/gemma-7b-it-lora | Gemma | — | free |
| @cf/ibm-granite/granite-4.0-h-micro | Granite | — | free |
| @cf/llava-hf/llava-1.5-7b-hf | LLaVA | — | free |
| @cf/meta-llama/llama-2-7b-chat-hf-lora | Llama | — | free |
| @cf/meta/llama-2-7b-chat-fp16 | Llama | — | free |
| @cf/meta/llama-2-7b-chat-int8 | Llama | — | free |
| @cf/meta/llama-3-8b-instruct | Llama | — | free |
| @cf/meta/llama-3-8b-instruct-awq | Llama | — | free |
| @cf/meta/llama-3.1-70b-instruct-fp8-fast | Llama | — | free |
| @cf/meta/llama-3.1-8b-instruct | Llama | — | free |
| @cf/meta/llama-3.1-8b-instruct-awq | Llama | — | free |
| @cf/meta/llama-3.1-8b-instruct-fp8 | Llama | — | free |
| @cf/meta/llama-3.1-8b-instruct-fp8-fast | Llama | — | free |
| @cf/meta/llama-3.2-11b-vision-instruct | Llama | — | free |
| @cf/meta/llama-3.2-1b-instruct | Llama | — | free |
| @cf/meta/llama-3.2-3b-instruct | Llama | — | free |
| @cf/meta/llama-3.3-70b-instruct-fp8-fast | Llama | — | free |
| @cf/meta/llama-guard-3-8b | Llama | — | free |
| @cf/mistral/mistral-7b-instruct-v0.1 | Mistral | — | free |
| @cf/mistral/mistral-7b-instruct-v0.2-lora | Mistral | — | free |
| @cf/mistralai/mistral-small-3.1-24b-instruct | Mistral | — | free |
| @cf/nousresearch/hermes-2-pro-mistral-7b | Mistral | — | free |
| @cf/nvidia/nemotron-3-120b-a12b | Nemotron | — | free |
| @cf/qwen/qwen3-30b-a3b-fp8 | Qwen | — | free |
| @cf/qwen/qwq-32b | Qwen | — | free |
| @hf/google/gemma-7b-it | Gemma | — | free |
| @hf/mistral/mistral-7b-instruct-v0.2 | Mistral | — | free |
| @hf/thebloke/llamaguard-7b-awq | Llama | — | free |
What you get
Edge AI near Workers
Quota/unit based
Model availability varies
Best with Cloudflare backend
Gotchas
No gotchas documented yet. Contribute one if you know a constraint we missed.
Sources
-
"Our free allocation allows anyone to use a total of 10,000 Neurons per day at no charge. To use more than 10,000 Neurons per day, you need to sign up for the Workers Paid plan... at $0.011 / 1,000 Neurons"
https://developers.cloudflare.com/workers-ai/platform/pricing/ -
"@cf/moonshotai/kimi-k2.6 | $0.950 per M input tokens ... $4.000 per M output tokens | 86364 neurons per M input tokens ... 363636 neurons per M output tokens"
https://developers.cloudflare.com/workers-ai/platform/pricing/ -
"Kimi K2.6 scores 54 on the Artificial Analysis Intelligence Index, placing it well above average among comparable models (averaging 30)."
https://artificialanalysis.ai/models/kimi-k2-6 -
"Text Generation: 300 requests per minute"
https://developers.cloudflare.com/workers-ai/platform/limits/ -
"We're excited to announce that our Workers AI inference platform is now Generally Available. After months of being in open beta, we've improved our service."
https://blog.cloudflare.com/workers-ai-ga-huggingface-loras-python-support/
Alternatives
Same category · similar quota
LLM7.io is a zero-friction gateway exposing 30+ models (DeepSeek, Gemini Flash, GPT-4o-mini, Qwen Coder, Mistral) through a single OpenAI-compatible endpoint with no registration for basic access.
