# Inference Providers credits show $0.00 despite no usage

**URL:** <https://discuss.huggingface.co/t/inference-providers-credits-show-0-00-despite-no-usage/190471>\
**Category:** Beginners\
**Created:** [October 13, 2026, 11:41am UTC](https://discuss.huggingface.co/t/inference-providers-credits-show-0-00-despite-no-usage/190471 "2026-10-13T11:41:33Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![1234jithu](https://avatars.discourse-cdn.com/v4/letter/1/7ea924/32.png) [@1234jithu](https://discuss.huggingface.co/u/1234jithu)\
**Post date:** [October 13, 2026, 11:41am UTC](https://discuss.huggingface.co/t/inference-providers-credits-show-0-00-despite-no-usage/190471/1 "2026-10-13T11:41:33Z")

</div>

Hello,

My Hugging Face account shows $0.00 in Inference Providers credits. My usage dashboard shows no inference requests and $0.00 usage. I have no payment method registered.

When I try to use `Qwen/Qwen2.5-1.5B-Instruct`, I receive an error saying I have no remaining credits.

Could you please clarify why my account has no available inference credits and whether my account is eligible for the free monthly allowance?

Thank you.

---

<div class="post-metadata">

**Author:** ![SalutefaceAAC](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/salutefaceaac/32/64306_2.png) [@SalutefaceAAC](https://discuss.huggingface.co/u/SalutefaceAAC)\
**Post date:** [October 13, 2026, 2:19pm UTC](https://discuss.huggingface.co/t/inference-providers-credits-show-0-00-despite-no-usage/190471/2 "2026-10-13T14:19:55Z")

</div>

Hi @1234jithu – this trips up a lot of people because HF has two separate inference systems with different billing:

**Inference Providers** (what you’re hitting): This is the paid API gateway to third-party providers (Fireworks, Novita, Together AI, etc.). It draws from your billing balance. Without a registered payment method, balance is $0.00 and requests fail.

**Serverless Inference API** (the free tier): This is HF’s own hosted inference, free for many smaller models with rate limits. It’s completely separate from Inference Providers.

**To use Qwen2.5-1.5B-Instruct for free** , use the serverless path – just omit the `provider` argument:

```python
from huggingface_hub import InferenceClient

client = InferenceClient(
    model="Qwen/Qwen2.5-1.5B-Instruct",
    token="hf_..." # your read token
)
result = client.text_generation("Hello!", max_new_tokens=100)
print(result)

```

No `provider=` means it routes to HF’s serverless API, not a paid third-party provider. Qwen2.5-1.5B is small enough to be available on the free tier.

You can also test right on the model card page – the **Inference API widget** on the right sidebar lets you send prompts without any code or billing setup.

If you eventually want to use Inference Providers for heavier workloads, add a payment method under **Settings / Billing**. But for this model size the serverless route works great.