NVIDIA NIM (build.nvidia.com)
permanent free · inference_host · verification: partially_verified (60%, doc_fetch, 2026-07-17) · setup guide →
The record
| Base URLs | https://integrate.api.nvidia.com/v1 openai-compatible |
|---|---|
| API key required | yes |
| Credit card | no |
| Phone verification | yes |
| May train on prompts | unknown |
| Quota | free_tier [rpm] “40 requests/minute (cheahjs); "Rate-limited (no daily token cap)" (mnfst); "~40 RPM" per model” |
| Signup friction | 5 steps · ~15m to first key |
| Env var | NVIDIA_API_KEY |
| Notes | Base URL confirmed from official API reference (https://integrate.api.nvidia.com, POST /v1/chat/completions across the LLM catalog). 100+ hosted models. Tool-calling support varies by model and was not confirmed. |
Models (4)
| model | capabilities | context | free | limits |
|---|---|---|---|---|
| deepseek-ai/deepseek-r1 | chat, reasoning | 128,000 | yes | 40 rpm |
| meta/llama-3.1-405b-instruct | chat | 128,000 | yes | 40 rpm |
| nvidia/llama-3.1-nemotron-ultra-253b-v1 | chat | 128,000 | yes | 40 rpm |
| qwen/qwen2.5-72b-instruct | chat | 128,000 | yes | 40 rpm |
Provenance
| source_nvidia_nim_docs | official_docs | trust: high | checked 2026-07-17 |
| source_repo_cheahjs | curated_repo | trust: medium_high | checked 2026-07-17 |
| source_repo_mnfst | curated_repo | trust: medium_high | checked 2026-07-17 |