Hugging Face Inference Endpoints

Functions 4.8/10

Dedicated autoscaling inference billed per minute of uptime. CPU from $0.032/core/hr; GPUs T4 $0.50/hr, L4 $0.80/hr, A100 and up to ~$45/hr; can scale to zero.

Hugging Face Inference Endpoints scores 4.8/10 for free value in Functions. The main constraint: Paid — From $0.032/CPU-core-hr; GPUs from $0.50/hr (T4). Dedicated autoscaling inference billed per minute of uptime. CPU from $0.032/core/hr; GPUs T4 $0.50/hr, L4 $0.80/hr, A100 and up to ~$45/hr; can scale to zero.

Category Functions Kind Paid plan

Free value score

4.8 / 10

Show the math

Overall 4.8/10

value = 4.8/100

weighted_components_v1

Invocation value 1/10

value = 0.01 ×

fixed_log_v1

Runtime fit 9/10

value = 9/100

policy_score_v1

Deploy friction 8/10

value = 8/100

policy_score_v1

Key metrics

$.032/coreh

T4 $0.50/hr

per-min

What you get

Paid — From $0.032/CPU-core-hr; GPUs from $0.50/hr (T4). Dedicated autoscaling inference billed per minute of uptime. CPU from $0.032/core/hr; GPUs T4 $0.50/hr, L4 $0.80/hr, A100 and up to ~$45/hr; can scale to zero.

~2 T4 GPU-hours per $1 ($0.50/hr); per-minute billing and scale-to-zero make it efficient for spiky inference. CPU endpoints extremely cheap for light

Gotchas

No gotchas documented yet. Contribute one if you know a constraint we missed.

Sources

"Inference Endpoints Dedicated inference, starting at $0.033/hour ... GPU Instances NVIDIA T4 ... $0.50 ... NVIDIA L4 ... $0.80"
https://huggingface.co/pricing
"As low as $0.06 per CPU core/hr and $0.6 per GPU/hr."
https://discuss.huggingface.co/t/misunderstanding-about-inference-endpoint-billing/57428

paid value serverless-functions

Open service ← All Functions

Alternatives

Same category · similar quota

Beam CloudBeam Cloud · paid · Paid plan

4.8

Pay-per-second (only while container runs): CPU $0.19/core/hr, RAM $0.0000056/GB/s, RTX 4090 $0.69/hr, A10G $1.05/hr, H100 $3.50/hr. Scale-to-zero; no charge for image pull or boot

4090 $192u/sH100 $972u/s

Checked 26d ago

No-sleepPaid only

Cloudflare ContainersCloudflare Containers · paid · Paid plan

4.8

Billed per 10ms active: 25 GiB-hours memory + 375 vCPU-minutes + 200 GB-hours disk included/month in the $5 Workers Paid plan; overage at listed rates. Scale-to-zero.

$20u/vCPU-s$2.5u/GiB-s25 GiB-h

Checked 26d ago

No-sleepPaid only

Cloudflare WorkersCloudflare Workers · paid · Paid plan

4.8

10M requests + 30M CPU-ms/month included; overage $0.30/M requests, $0.02/M CPU-ms. No duration (I/O wait) charge, no egress, edge in 300+ cities.

$5/mo min10M req30M CPU-ms

Checked 26d ago

Best valueLow/no egressPaid only