Dear NVIDIA Support Team,
I am writing to request a rate limit increase for my NVIDIA NIM API account for personal development and model evaluation purposes.
Current Limit: 40 RPM
Requested Limit: 200 RPM
Background:
I am an AI enthusiast based in Taiwan, actively evaluating NVIDIA NIM models for personal learning and development. I have been using deepseek-ai/deepseek-v4-flash and nvidia/nemotron-3-super-120b-a12b through the NIM API, comparing their performance across different tasks including reasoning, coding assistance, and structured output generation.
Why I Need the Increase:
My evaluation workflow requires sending multiple concurrent inference requests to:
- Benchmark model response times under different prompt patterns
- Compare output quality across models for the same task
- Test edge cases with structured parameter variations
- Validate consistency and reliability across repeated queries
With the current 40 RPM limit, I am frequently interrupted by 429 errors during these evaluation sessions, which prevents me from gathering meaningful benchmark data. The intermittent rate limiting makes it difficult to distinguish between genuine model behavior and throttling artifacts in my test results.
My Setup:
- Models tested: deepseek-ai/deepseek-v4-flash, nvidia/nemotron-3-super-120b-a12b
- Environment: Self-hosted Linux server (Oracle Cloud ARM instance)
- Usage: Personal, non-commercial model evaluation and skill development
- Testing period: Ongoing
Commitment to Fair Use:
I fully understand and respect NVIDIA’s fair use policy for the free tier. This is strictly for personal evaluation and learning — I am not running a production service or commercial application. An increase to 200 RPM would allow me to complete meaningful evaluation sessions without constant interruptions.
I would also like to mention that a positive experience with the free tier would strongly influence my decision to adopt paid NVIDIA solutions (such as DGX Cloud or Brev.dev) for future projects that require higher throughput.
Thank you for considering my request. I appreciate NVIDIA’s commitment to providing free, high-quality inference to the developer community.
Best regards,
Ninelive Hu
Taiwan