The End of NowLLM : Closing an In-House Era

[Its_Azar]

ServiceNow is ending the default use of its in-house large language model, Now LLM, and shifting towards integrating advanced third-party models like Google Gemini and OpenAI GPT. This update shows ServiceNow's strategy to provide more powerful, flexible, and cutting-edge AI capabilities for its users. well I think that it's for the greater good lol.

 

Key Updates:

New Third-Party Models
With the July 2026 release, ServiceNow introduced several third-party LLMs for Now Assist and related services:

• Google Gemini 3.5 Flash replaces the earlier Gemini 2.5 Flash.
• OpenAI GPT 5.1 and OpenAI GPT 5.4 Mini are added to the platform.

These models now form the core of ServiceNow’s generative AI offerings, providing users with improved performance, larger knowledge bases, and enhanced contextual understanding.

Now LLM Is No Longer the Default Model
The July update marked a significant shift the Now LLM is no longer the default model for any AI skills or agents on the platform. This means existing users must:

• Update their AI configurations
• Choose from available third-party LLMs as their new default
• basically, change all thing that previously were consuming Now LLM

Retirement of Now LLM Long-Term Support (LTS) SKU
As of June 2026, the Now LLM LTS SKU has been retired, signaling an end to long-term support for the in-house model. Customers currently using this SKU will need to migrate to third-party models to ensure continued support and compatibility.

Why This Shift?

Access to Leading-Edge AI Tech
Third-party models like OpenAI and Google Gemini are industry leaders known for pushing the boundaries of AI, creativity, and language understanding. Integrating them into ServiceNow means users benefit from:

• More accurate responses
• Enhanced contextual understanding
• Frequent model updates and improvements
• Broader language and knowledge coverage

Simplified AI Management
By adopting third-party providers as defaults, ServiceNow reduces internal overhead for maintaining large AI models, focusing instead on seamless integration and user experience enhancements.

What Users Need to Do Next

If you are currently using Now LLM, here are your recommended next steps:

1. Evaluate Third-Party Models
   Test and compare Google Gemini 3.5 Flash, OpenAI GPT 5.1, and GPT 5.4 Mini to find the best fit for your business needs.

2. Update AI Configurations
   Change your generative AI configuration records to set a new default model.

3. Plan Migration from Now LLM LTS
   Since the long-term support model is retired, start planning your migration to avoid disruptions.

Bye until next time.

[sasikumarpk]

What about OOTB AI capabilities like case summarisation, suggested next steps, knowledge gaps, AI search, knowledge article generation, etc.? How would we handle these OOTB use cases where there is no need for any external LLMs 

[Its_Azar]

Hi @sasikumarpk 

Those will still be available but the only change is instead of nowLLM now we have gemini and gpt modules to choose and if you already have an usecase which is consuming the nowLLM then i would suggest to change the model.

[Vladka D_]

Great update. One question: what does this mean from a consumption and cost perspective? If customers move from Now LLM to GPT or Gemini models, will AI consumption, credits, or licensing remain unchanged, or will different models have different consumption rates/cost implications? This seems like an important consideration for customers planning their migration.

[alexpanzare]

RE: @Vladka D_ 

My understanding is consumption stays the same. Usage is measured in assists, and the assist rate is identical across Azure OpenAI, AWS Claude, Google Gemini, and Now LLM Service. There is no per model pricing tier.

ServiceNow holds the provider contracts and licensing. No separate agreements or API keys needed for the integrated option.

Sharing this as directional guidance. Pricing and entitlement questions should go through your account team