As NetSuite continues to expand its generative AI capabilities, developers working with the N/llm module must understand the usage limits, monthly quotas, and concurrency rules that apply to LLM-related SuiteScript calls.

If you’re building AI-driven Suitelets, Map/Reduce scripts, or integrations using the N/llm module, ignoring these limits can lead to unexpected runtime errors such as:

• “Exceeded N/llm usage limit”
• “Concurrent LLM calls exceed allowed threshold”

This guide explains everything you need to know about:

✔ Monthly AI usage pools
✔ How SuiteApps receive their own usage quota
✔ How to increase usage with OCI
✔ Concurrency limits for generate and embed methods
✔ Practical best-practices to avoid hitting these limits

Let’s break it down.

---

⭐ 1. Usage Limits for N/llm Methods

NetSuite gives every account access to a free monthly pool of generative AI usage.

This pool is used for:

👉 Generate Methods

• llm.generateText()
• llm.generateTextStreamed()
• llm.evaluatePrompt()
• Their .promise() versions

👉 Embed Methods

• llm.embed()
• llm.embed.promise()

Each successful response from the LLM deducts usage from your monthly quota.

🎯 Key Notes About Usage Limits

• Generate and embed usage are tracked separately
• Unused monthly quota does not carry over
• Usage resets at the beginning of each month
• You can view usage on the AI Preferences page

This makes it easy to monitor how your scripts are consuming AI resources.

---

⭐ 2. SuiteApps Have Their Own Usage Pools

This is one of the most important — and often overlooked — parts of NetSuite’s design.

✔ Each SuiteApp installed in your account gets its own separate usage pool.

✔ These pools are independent from your main account pool.
✔ SuiteApps cannot drain your main account’s AI quota.

Example:

If your account has:

• Your internal scripts
• SuiteApp A
• SuiteApp B

Then you get:

• 1 main usage pool (for your scripts)
• 1 pool for SuiteApp A
• 1 separate pool for SuiteApp B

So technically you have 3 separate monthly usage allocations.

Additional SuiteApp Details

• Generate vs embed usage are tracked separately within each SuiteApp
• Usage numbers for SuiteApps do not appear on the AI Preferences page
• A SuiteApp developer must provide its own usage dashboard

This distribution prevents a SuiteApp from consuming your entire quota and blocking your internal automations.

---

⭐ 3. How to Get More Usage (Using Your OCI Account)

If your monthly free pool isn’t enough, you can add:

✔ Your Own OCI Credentials

Once configured, NetSuite will:

• Stop using the free monthly pool
• Consume usage from your OCI Generative AI service instead

This is ideal for:

• Large enterprise workloads
• Scripts that handle large prompt volumes
• AI-heavy SuiteApps
• High-frequency Map/Reduce automations

You must configure OCI credentials either:

• In your SuiteScript code (via ociConfig object), or
• On the AI Preferences → Settings tab

Once done, the usage becomes effectively unlimited depending on your OCI plan.

---

⭐ 4. Concurrency Limits for N/llm Methods

In addition to monthly usage, NetSuite enforces concurrency limits to prevent too many simultaneous LLM calls.

These limits apply at runtime.

---

🧠 Generate Methods — Maximum 5 Concurrent Calls

Applies to:

• llm.generateText()
• llm.generateText.promise()
• llm.generateTextStreamed()
• llm.generateTextStreamed.promise()
• llm.evaluatePrompt()
• llm.evaluatePrompt.promise()
• llm.evaluatePromptStreamed()
• llm.evaluatePromptStreamed.promise()

Limit:

➡ Up to 5 concurrent requests

If you exceed this, NetSuite will throw a concurrency error.

---

🧠 Embed Methods — Maximum 5 Concurrent Calls

Applies to:

• llm.embed()
• llm.embed.promise()

Limit:

➡ Up to 5 concurrent embed operations

Generate and embed concurrency are tracked separately, so you could theoretically run:

• 5 generate operations
• AND 5 embed operations
  at the same time

Without hitting concurrency caps.

---

⭐ 5. Practical Tips to Avoid Usage & Concurrency Errors

Here are best practices for real-world NetSuite development:

---

✔ 1. Queue or batch multiple AI requests

Instead of firing 10 calls at once, stagger them using:

• Scheduled scripts
• Async queues
• Delayed promise chains

---

✔ 2. Use the streaming methods wisely

Streamed responses count the same as non-streamed but may help reduce failures.

---

✔ 3. Use Prompt Studio for reusable prompts

It reduces prompt size, lowering token usage and cost.

---

✔ 4. Cache results when possible

If you’re using AI to classify or summarize similar data, store results.

---

✔ 5. Switch to OCI for enterprise workloads

If you expect high traffic:

• Enable OCI config
• Bypass NetSuite’s free limits
• Avoid end-of-month failures

---

✔ 6. Monitor usage frequently

Use the AI Preferences page to avoid surprises near month-end.

---

⭐ 6. Final Thoughts

As more developers build AI-powered scripts, understanding usage and concurrency limits is essential to preventing failures and scaling your automations safely.

The N/llm module is incredibly powerful — but like all AI systems, it must be used responsibly and optimized to avoid:

• Concurrency overload
• Monthly quota exhaustion
• Unexpected errors in production

With proper planning, you can use NetSuite’s generative AI APIs efficiently and reliably across your entire automation stack.