Request to enable "Public API Endpoints" — models listed in /v1/models return 404 "Not found for account"

AI & Data Science NVIDIA NIM Access/Accounts
, , , , ,
1

Hello NVIDIA team,

I’m requesting the “Public API Endpoints” permission for my personal organization.

Several models are returned by GET /v1/models but fail with HTTP 404 “Not found for account” when called via POST /v1/chat/completions.

ACCOUNT DETAILS

  • Email: erick@eaclog.com
  • API key prefix: nvapi-urRzsm-
  • Account ID: mgOSo549FLoG1eCYyX-E4o_GFBedKbjJWXIG9aaH_u0
  • Organization: personal

AFFECTED MODELS (all listed by GET /v1/models, all 404 on inference)

  1. moonshotai/kimi-k2.6
    {“status”:404,“title”:“Not Found”,“detail”:“Function ‘23d4f03a-b8a6-4adb-a183-7daa083a09cc’: Not found for account ‘mgOSo549FLoG1eCYyX-E4o_GFBedKbjJWXIG9aaH_u0’”}

  2. writer/palmyra-fin-70b-32k
    {“status”:404,“title”:“Not Found”,“detail”:“Function ‘316490c6-f1ed-41f9-9da8-3fa9e885653b’: Not found for account ‘mgOSo549FLoG1eCYyX-E4o_GFBedKbjJWXIG9aaH_u0’”}

  3. mistralai/mistral-large-2-instruct
    {“status”:404,“title”:“Not Found”,“detail”:“Function ‘7fadd4de-e22a-48e4-90e9-f02ef14a74b9’: Not found for account ‘mgOSo549FLoG1eCYyX-E4o_GFBedKbjJWXIG9aaH_u0’”}

WHAT WORKS WITH THE SAME KEY (so this is not an authentication problem)

  • nvidia/nemotron-3-ultra-550b-a55b → 200 OK
  • nvidia/nemotron-3-super-120b-a12b → 200 OK
  • nvidia/nemotron-3-nano-30b-a3b → 200 OK
  • nvidia/llama-3.3-nemotron-super-49b-v1.5 → 200 OK
  • z-ai/glm-5.2 → 200 OK
  • deepseek-ai/deepseek-v4-pro → 200 OK
  • minimaxai/minimax-m3 → 200 OK
  • meta/llama-3.1-70b-instruct → 200 OK

REPRODUCTION

curl https://integrate.api.nvidia.com/v1/chat/completions
-H “Authorization: Bearer $NVIDIA_API_KEY”
-H “Content-Type: application/json”
-d ‘{“model”:“moonshotai/kimi-k2.6”,“messages”:[{“role”:“user”,“content”:“ok”}],“max_tokens”:4}’

The same request against nvidia/nemotron-3-super-120b-a12b returns 200.

NOTE

The Function ID for moonshotai/kimi-k2.6 in my error (‘23d4f03a-…’) is identical to the one reported by other users in this forum, which suggests the function ID is per-model and the gate is on the account permission, not on the model deployment.

USE CASE

Commercial supply-chain platform (China → Brazil imports). We run a fleet of specialist agents; kimi-k2.6 is needed for the Chinese-language sourcing agent, and palmyra-fin-70b-32k for the tax/customs agent.

Thank you.