Developer hub
NVIDIA NIM
Model prices vs the direct API
build.nvidia.comUnverified
- Fee on tokens
- Not published
- Hosted API endpoints free for NVIDIA Developer Program prototyping; production self-hosted NIM requires an NVIDIA AI Enterprise license from $4,500 per GPU per year (about $1 per GPU per hour in the cloud); no per-token price
- BYOK fee
- No BYOK
- Entry plan
- Usage based
- Hosting
- Hosted
Features
- Fallbacks
- Load balancing
- Caching
- Guardrails
- Observability
- Rate limits
- Budgets
- Prompts
- OpenAI-compatible
- Bring your own key
- Data residency
- Private networking
What it costs on real workloads
Claude Sonnet 5n/avs $400.00 direct
GPT-6 Soln/avs $400.00 direct
Gemini 3.8 Flashn/avs $150.00 direct
DeepSeek V4.1 Flashn/avs $27.00 direct
100M input and 20M output tokens a month, paying with gateway credits.
The terms, in their words
NVIDIA's API catalog of hosted model endpoints for prototyping plus downloadable NIM inference microservices for self-hosting on NVIDIA GPUs.
- Markup
- Hosted API endpoints free for NVIDIA Developer Program prototyping; production self-hosted NIM requires an NVIDIA AI Enterprise license from $4,500 per GPU per year (about $1 per GPU per hour in the cloud); no per-token price
- Free tier
- Free inference endpoints for NVIDIA Developer Program members (prototyping only); 90-day NVIDIA AI Enterprise trial
- These licenses start at $4500 per GPU per year or ~ $1 per GPU per hour in the cloud. Pricing is based on the number of GPUs, not the number of NIMs.
- Members of the NVIDIA Developer Program have free access to NIM API endpoints for prototyping
- Using NIM in production requires an NVIDIA AI Enterprise license.
NVIDIA NIM vs
Hugging Face Inference
Cloudflare Workers AI
Helicone
Kong AI Gateway
LiteLLM
Vercel AI GatewayPricing as read from the gateway's own pages on the date shown; your contract may differ. Logos via logo.dev; trademarks belong to their owners.