NVIDIA NIM Rate Limit Increase Request: 40 RPM to 100/200 RPM

AI & Data Science NVIDIA NIM Access/Accounts
,
1

Hello NVIDIA Support Team,

I am requesting a rate limit increase for my NVIDIA NIM API account.

Account Details:

  • Name: Ofek Grunfeld
  • Current Limit: 40 RPM
  • Requested Limit: 100 RPM initially, with consideration for 200 RPM if available
  • Usage Type: Personal development, research, testing, and evaluation
  • Primary Models: moonshotai/kimi-k2.5, qwen/qwen3.5-397b-a17b, z-ai/glm5, and NVIDIA-hosted speech/ASR models where available

Reason for Request:

I am building and testing a personal AI development workspace that uses NVIDIA NIM as the inference backend for agentic software engineering, knowledge
workflows, and media-processing prototypes.

My main use cases are:

  1. Agentic software development
    I use NIM-backed coding agents for iterative development tasks such as code generation, debugging, refactoring, test analysis, documentation, and tool/
    plugin experimentation. These workflows often require many small sequential requests, and sometimes several concurrent requests, especially when testing
    tool-calling behavior or comparing model outputs.

  2. Obsidian-based knowledge and productivity system
    I am developing a personal knowledge-management workflow around Obsidian on both desktop and mobile. The goal is to connect notes, tasks, research,
    planning, and coding context into a structured AI-assisted system. This is still personal development and evaluation, but realistic testing requires
    frequent short model calls for summarization, classification, task extraction, and context organization.

  3. Translator and subtitle-captioning prototype
    I am also testing a workflow that combines speech recognition with LLM-based translation, cleanup, subtitle formatting, and summarization. Even a single
    media file can require multiple calls because audio must be chunked, transcribed, translated, reviewed, and reformatted. The current 40 RPM limit is
    quickly reached during normal testing of chunk sizes, prompt variants, and multilingual output quality.

Why the Current Limit Is Restrictive:

The 40 RPM limit is enough for simple manual prompts, but it is too low for realistic testing of agentic and multi-step workflows. I frequently encounter
429 errors during normal development sessions, especially when testing concurrent tool calls, subtitle/transcription pipelines, and Obsidian note-
processing flows.

I am already using a local rate limiter and conservative concurrency settings to avoid unnecessary load. I am not using multiple accounts or key rotation
to bypass limits. I am also not using NIM for production traffic, resale, public user-facing services, scraping, or abuse. This request is for legitimate
personal development, prototyping, research, and learning.

Requested Increase:

An increase to 100 RPM would significantly improve my ability to test these workflows reliably. If possible, 200 RPM would be ideal for bursty local
development sessions involving agentic coding and transcription/translation pipelines.

I am happy to provide my NVIDIA Cloud Account ID, additional details, or usage examples if needed.

Thank you for making NVIDIA NIM available to developers. It has been very useful for evaluating modern AI workflows and building practical prototypes.

Best regards,
Ofek Grunfeld

2

Just to clarify, this is only for personal development and testing, not production. Agentic coding creates many small requests, so 40 RPM gets hit quickly
even with local rate limiting. A 100 RPM limit would already help a lot. Thanks!