# Use LLM block returning null with reasoning models

**URL:** <https://community.palantir.com/t/use-llm-block-returning-null-with-reasoning-models/3575>\
**Category:** Ask the Community\
**Tags:** pipeline-builder\
**Created:** [April 17, 2025, 7:29pm UTC](https://community.palantir.com/t/use-llm-block-returning-null-with-reasoning-models/3575 "2025-04-17T19:29:30Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![tanmayb](https://avatars.discourse-cdn.com/v4/letter/t/dfb087/32.png) [@tanmayb](https://community.palantir.com/u/tanmayb)\
**Post date:** [April 17, 2025, 7:29pm UTC](https://community.palantir.com/t/use-llm-block-returning-null-with-reasoning-models/3575/1 "2025-04-17T19:29:30Z")

</div>

I have a simple test prompt, as shown below. It works when using most models, but largely returns “null” on using reasoning-based models. o3-mini and o1-mini always returned null. Grok-3-Mini-Reasoning returned null for all but one.

 ![Screenshot 2025-04-17 at 3.25.11 PM](https://canada1.discourse-cdn.com/palantir/original/2X/7/7d554aa02929812406c74eee3a7c5b4797d2ff9f.png)

 ![Screenshot 2025-04-17 at 3.25.51 PM](https://canada1.discourse-cdn.com/palantir/original/2X/f/feba0d720f748dd26652111cbb07b35dc46f44ea.png)

Looking at the raw prompts, it doesn’t seem like there’s an output/try at all.

 ![Screenshot 2025-04-17 at 3.28.25 PM](https://canada1.discourse-cdn.com/palantir/original/2X/6/66993c9ab1f9273a949dea4939a0c76288296b37.png)

Am I missing any config inputs? How do I debug this? Thanks.

---

<div class="post-metadata">

**Author:** ![tanmayb](https://avatars.discourse-cdn.com/v4/letter/t/dfb087/32.png) [@tanmayb](https://community.palantir.com/u/tanmayb)\
**Post date:** [April 17, 2025, 7:37pm UTC](https://community.palantir.com/t/use-llm-block-returning-null-with-reasoning-models/3575/2 "2025-04-17T19:37:42Z")

</div>

Turned on errors and this is the error I’m getting:

```auto
{"ok":null,"error":"Encountered an unknown exception during completion"}

```

Grok-3-Mini-Reasoning does return non-null outputs for all test inputs now.

---

<div class="post-metadata">

**Author:** ![sperchanok](https://avatars.discourse-cdn.com/v4/letter/s/b5a626/32.png) [@sperchanok](https://community.palantir.com/u/sperchanok)\
**Post date:** [April 17, 2025, 8:36pm UTC](https://community.palantir.com/t/use-llm-block-returning-null-with-reasoning-models/3575/3 "2025-04-17T20:36:44Z")

</div>

Hello! I did a little digging and this appears to be an error with our language model services. Working to resolve with other internal dev teams.

However, this is a good flag that we may need to show better error messages in this case.

---

<div class="post-metadata">

**Author:** ![system](https://yyz1.discourse-cdn.com/palantir/user_avatar/community.palantir.com/system/32/3_2.png) [@system](https://community.palantir.com/u/system)\
**Post date:** [June 16, 2025, 8:37pm UTC](https://community.palantir.com/t/use-llm-block-returning-null-with-reasoning-models/3575/4 "2025-06-16T20:37:21Z")

</div>

Closed on June 16, 2025, 8:37pm UTC

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.
