Hello NVIDIA Support Team,
I’m a solo developer using NVIDIA NIM for personal learning and development. After a systematic test of the /v1/chat/completions endpoint with my API key, I found that NVIDIA’s own models and some third-party models work correctly (HTTP 200), but several third-party models return 404 “Function not found for account”, even though they are listed in /v1/models.
Models that work (HTTP 200):
nvidia/nemotron-3-ultra-550b-a55b
mistralai/mistral-large-4-1t-a52b
poolside/laguna-xs-2.1
nvidia/nemotron-3.5-lightning-30b-a3b
meta/muse-glimmer-30b
Models that return 404 “Function not found for account”:
moonshotai/kimi-k2.6
deepseek-ai/deepseek-coder-6.7b-instruct
z-ai/glm-5.3-flash (also 000/timeout)
openai/gpt-oss-20b (also 000/timeout)
mistralai/mistral-large-2-instruct
meta/codellama-70b
ibm/granite-34b-code-instruct
and others
This looks like a per-account entitlement issue: the models are in the public catalog but are not provisioned for my account. Could you please enable access for my organization to the third-party models on the free tier, specifically moonshotai/kimi-k2.6 (coding) and z-ai/glm-5.3-flash (fast inference)?
Account details:Registered Email: leokol045@gmail.com
Account ID: 6GtE_0ErIaGLJZSKBuQcqdoVGTKQKUMALAJH2BW9EVo
Intended Use: Personal development and learning (non-commercial)