# Meta-Llama-Guard-3-8B doesn't work via API

**URL:** <https://community.sambanova.ai/t/meta-llama-guard-3-8b-doesnt-work-via-api/739>\
**Category:** Community Feedback\
**Tags:** jira-issue\
**Created:** [December 9, 2024, 11:24pm UTC](https://community.sambanova.ai/t/meta-llama-guard-3-8b-doesnt-work-via-api/739 "2024-12-09T23:24:07Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![qdrddr](https://sea1.discourse-cdn.com/sambanova/user_avatar/community.sambanova.ai/qdrddr/32/1506_2.png) [@qdrddr](https://community.sambanova.ai/u/qdrddr)\
**Post date:** [December 9, 2024, 11:24pm UTC](https://community.sambanova.ai/t/meta-llama-guard-3-8b-doesnt-work-via-api/739/1 "2024-12-09T23:24:07Z")

</div>

When I use Playground via UI ([https://cloud.sambanova.ai](https://cloud.sambanova.ai)), Meta-Llama-Guard-3-8B model works just fine. However, if I use it via API to access the same model, I always get “Something went wrong!”:

```shell
curl -H "Authorization: Bearer $API_KEY" \
     -H "Content-Type: application/json" \
     -d '{
	"stream": true,
	"model": "Meta-Llama-Guard-3-8B",
	"messages": [
		{
			"role": "system",
			"content": "You are a helpful assistant"
		},
		{
			"role": "user",
			"content": "Hello"
		}
	]
	}' \
     -X POST https://api.sambanova.ai/v1/chat/completions
#Something went wrong!

```

While other models work fine via API.

---

<div class="post-metadata">

**Author:** ![omkar.gangan](https://sea1.discourse-cdn.com/sambanova/user_avatar/community.sambanova.ai/omkar.gangan/32/217_2.png) [@omkar.gangan](https://community.sambanova.ai/u/omkar.gangan)\
**Post date:** [December 10, 2024, 4:06am UTC](https://community.sambanova.ai/t/meta-llama-guard-3-8b-doesnt-work-via-api/739/2 "2024-12-10T04:06:46Z")

</div>

Hello @qdrddr, Welcome to the community!  
Meta-Llama-Guard-3-8B model currently does not support the system prompt.  
We are working on it. Please remove system prompt from the curl and try it again.

```auto
curl -H "Authorization: Bearer $API_KEY" \
     -H "Content-Type: application/json" \
     -d '{
	"stream": true,
	"model": "Meta-Llama-Guard-3-8B",
	"messages": [
		{
			"role": "user",
			"content": "Hello"
		}
	]
	}' \
     -X POST https://api.sambanova.ai/v1/chat/completions

```

Thanks & Regards

---

<div class="post-metadata">

**Author:** ![qdrddr](https://sea1.discourse-cdn.com/sambanova/user_avatar/community.sambanova.ai/qdrddr/32/1506_2.png) [@qdrddr](https://community.sambanova.ai/u/qdrddr)\
**Post date:** [December 10, 2024, 2:15pm UTC](https://community.sambanova.ai/t/meta-llama-guard-3-8b-doesnt-work-via-api/739/4 "2024-12-10T14:15:09Z")

</div>

Thanks, that worked! @omkar.gangan  
Please make sure to remove the system prompt on this page in the example for the Meta-Llama-Guard-3-8B:

> **[SambaNova Cloud](https://cloud.sambanova.ai/apis)**
>
> Preview AI-enabled Fastest Inference APIs in the world.

 ![CleanShot 2024-12-10 at 08.14.48](https://us1.discourse-cdn.com/sambanova/original/2X/6/66c6c539a67fe2feeafc8af1770e975664334c5e.png)

---

<div class="post-metadata">

**Author:** ![Coby](https://sea1.discourse-cdn.com/sambanova/user_avatar/community.sambanova.ai/coby/32/17_2.png) [@Coby](https://community.sambanova.ai/u/Coby)\
**Post date:** [December 21, 2024, 9:45am UTC](https://community.sambanova.ai/t/meta-llama-guard-3-8b-doesnt-work-via-api/739/5 "2024-12-21T09:45:23Z")

</div>

@qdrddr I will file a Jira to get that tended to thank you for pointing it out