Price checkPer million tokens
Developer (Free credits)FreePay-as-you-go (hosted NIM endpoints)Usage-basedEnterprise (AI Enterprise license + DGX Cloud)Custom
See all 3 plans

All NVIDIA NIM Plans & Pricing

Plan Monthly Annual Best For
View all features by plan (compare side-by-side)

Developer (Free credits)

  • Free API credits on build.nvidia.com
  • Access to 50+ hosted models
  • NVIDIA, Meta, Mistral, Microsoft open models

Pay-as-you-go (hosted NIM endpoints)

  • Llama 3.3 70B: $0.90/1M tokens blended
  • Mixtral 8x22B: $1.20/1M
  • Nemotron 70B: $0.90/1M
  • Hosted behind NVIDIA inference infra

Enterprise (AI Enterprise license + DGX Cloud)

  • AI Enterprise subscription
  • DGX Cloud dedicated capacity
  • On-prem deployment support
Pricing Alerts

Track NVIDIA NIM pricing

Get an email when NVIDIA NIM's pricing changes — plus the weekly SaaS Price Watch: verified price changes and deals across 3,000+ products. One-click unsubscribe.

Try NVIDIA NIM Free
Compare NVIDIA NIM with alternativesAdjust seats, lock a tier, add up to 2 more products side-by-side. Shareable URL.
Quick Answer
Last verified:
Medium confidence

NVIDIA NIM offers usage-based pricing from $0.900–$1.20 per million tokens as of October 2026 and custom pricing for larger requirements. Plans: Developer (Free credits) (free), and Pay-as-you-go (hosted NIM endpoints) (usage-based). Custom pricing is available on request. Usage cost depends on the selected model and generation volume.

Use the interactive pricing calculator to estimate your exact cost based on team size and requirements.

  • Free tier: Yes

NVIDIA NIM offers 3 pricing tiers: Developer (Free credits), Pay-as-you-go (hosted NIM endpoints), Enterprise (AI Enterprise license + DGX Cloud). A free plan is available. Paid plans include Pay-as-you-go (hosted NIM endpoints) (usage-based). The Pay-as-you-go (hosted NIM endpoints) plan is production inference on open models.

Compared to other llm api providers software, NVIDIA NIM is positioned at the budget-friendly price point.

  • 4 documented hidden costs beyond list price
Last verified:

How much does NVIDIA NIM cost?

NVIDIA NIM offers a free plan plus usage-based pricing from $0.900–$1.20 per million tokens, with custom pricing available for larger requirements. Plans include Developer (Free credits) (free), Pay-as-you-go (hosted NIM endpoints) (usage-based), Enterprise (AI Enterprise license + DGX Cloud) (custom pricing).

NVIDIA NIM Pricing Overview

NVIDIA NIM offers usage-based pricing from $0.900–$1.20 per million tokens and custom pricing for larger requirements. The Developer (Free credits) plan is free and is best for prototyping and evaluation. The Pay-as-you-go (hosted NIM endpoints) plan is usage-based and is designed for production inference on open models. The Enterprise (AI Enterprise license + DGX Cloud) plan requires contacting sales for a custom quote and is designed for regulated enterprise inference workloads.

There are at least 4 documented hidden costs beyond NVIDIA NIM's list price, including implementation, training, and add-on fees.

This pricing was last verified in August 19, 2026.

Try NVIDIA NIM Free

NVIDIA NIM pricing starts at $0/month on the Developer (Free credits) plan, giving developers API access to hosted NIM microservice endpoints. The Pay-as-you-go (hosted NIM endpoints) tier charges per token consumed, with hosted model rates on OpenRouter ranging from $0.04–$1.20 per million input tokens depending on model size. Organizations requiring enterprise deployment on DGX Cloud or on-premises infrastructure are served by the Enterprise (AI Enterprise license + DGX Cloud) plan, which is custom-quoted.

How NVIDIA NIM Pricing Compares

Compare NVIDIA NIM pricing against top alternatives in LLM API Providers.

Groq $0-$4.0 per million tokens Compare → Together AI $0.03-$9.95 per million tokens / hour Compare → Fireworks AI $0-$9 per million tokens / hour Compare →

Usage-Based Rates

Per-unit pricing for NVIDIA NIM API usage.

Pay-as-you-go (hosted NIM endpoints)

Model Unit Rate
llama-3-3-70b-nim 131072 ctx 1M tokens $0.900
mixtral-8x22b-nim 65536 ctx 1M tokens $1.20
nemotron-70b 131072 ctx 1M tokens $0.900
  • NIM = NVIDIA Inference Microservice. Same pricing on cloud-hosted + on-prem DGX deployments.

Compare NVIDIA NIM vs Alternatives

Before committing to NVIDIA NIM, compare pricing with these 3 alternatives in the same category.

VSGroq
Free

Prototyping and evaluation

Full comparison
VSTogether AI
From $0.03 per million tokens / hour

Variable-volume API usage

Full comparison
VSFireworks AI
Free

Variable-volume API usage

Full comparison

All NVIDIA NIM alternatives & migration guides

What Companies Actually Pay for NVIDIA NIM

Median per-1M-token pricing across 3 models
Input $0.085/1M
Output $0.400/1M
Flagship models in this provider's catalog
Model Input /1M Output /1M Blended /1M
NVIDIA: Nemotron 3 Ultra (nvidia/nemotron-3-ultra-550b-a55b) $0.600 $3.60 $0.164
NVIDIA: Nemotron 3 Super (nvidia/nemotron-3-super-120b-a12b) $0.085 $0.400 $0.164
NVIDIA: Nemotron 3 Nano 30B A3B (nvidia/nemotron-3-nano-30b-a3b) $0.050 $0.200 $0.164
Review scores
Third-party review aggregates, as of Aug 2026
Source: OpenRouter API — medians aggregated from 3 models routed. Reflects router-surface pricing (may include modest markup vs direct provider rates).

NVIDIA NIM Year 1 Total Cost by Company Size

Real deployment costs including licenses, implementation, training, and admin — not just the sticker price.

Developer Evaluation $0 Year 1 total

Individual developer exploring NVIDIA NIM hosted endpoints using the free credits tier for prototyping and evaluation

Small team at median token pricing $5 Year 1 total
$62.40/year
Total $5

Team consuming 10M input tokens and 10M output tokens per month via hosted NIM endpoints at provider median pricing ($0.095/1M input, $0.425/1M output)

Production app on Llama 3.1 Nemotron 70B $120 Year 1 total
$1,440/year
Total $120

Production application processing 50M input tokens and 50M output tokens per month via the highest-priced NVIDIA NIM model at $1.20/1M input and $1.20/1M output

Pay-as-you-go API Usage via OpenRouter Median $0.24/1M input tokens, $0.425/1M output tokens (via OpenRouter) Year 1 total
via OpenRouter
Total Median $0.24/1M input tokens, $0.425/1M output tokens (via OpenRouter)

A team consuming NVIDIA Nemotron models via OpenRouter's hosted NIM endpoints on a usage basis. Based on provider median pricing of $0.24/1M input tokens and $0.425/1M output tokens across 4 sampled models.

Enterprise Deployment (Custom Pricing) Custom-quoted (contact NVIDIA) Year 1 total
contact NVIDIA
Total Custom-quoted (contact NVIDIA)

An enterprise deploying NVIDIA NIM microservices on-premises or via DGX Cloud with an AI Enterprise license, requiring production SLAs and dedicated infrastructure.

CURRENT TIER DATA

How NVIDIA NIM Pricing Compares

Software Starting Price Top Price
NVIDIA NIM $0.10 per million tokens $10 per million tokens
Amazon Bedrock $0.07 per million tokens $75 per million tokens
Anyscale Free $4.9591 per million tokens
Baidu ERNIE API $0.1 per million tokens $10 per million tokens
Cerebras Inference API $0.1 per million tokens $6 per million tokens
Claude API $0.1 per million tokens $50 per million tokens

Detailed pricing comparisons:

Browse all LLM API Providers pricing →

4 NVIDIA NIM Hidden Costs Beyond the List Price

Beyond the listed price, NVIDIA NIM has at least 4 documented hidden costs that can significantly increase total cost of ownership.

Watch for 4 hidden costs
  • Storage Costs $100-$200
    medium 1 source
    Grounded summary Storage Costs: Storing large models like Llama 2 70B (140GB unquantized, 35GB quantized) on cloud storage can cost $100-$200 monthly, depending on the provider and redundancy requirements.
  • Container Registry & Image Management $20-$50
    low 1 source
    Grounded summary Container Registry and Image Management: Pulling and caching NIM container images, which can be 30-50GB for large models, incurs average container registry costs of $20-$50 monthly for production-grade storage and bandwidth.
  • Kubernetes Infrastructure Overhead
    high 1 source
    Grounded summary Kubernetes Infrastructure Overhead (Self-Hosting): Running a production Kubernetes cluster for self-hosting NIM involves costs such as $300-$500 monthly for master node(s), $100-$300 monthly for monitoring infrastructure, $200-$500 monthly for log...
  • Idle GPU Costs $1,950
    high 1 source
    Grounded summary For example, 50 real-time requests daily processing 100k tokens each (5 million daily tokens) can cost $1,950 for a GPU that is 99% idle, whereas an equivalent API cost might be $23.
Tip

Ask your NVIDIA NIM sales rep about these costs upfront. Getting them in writing before signing can save you from surprise charges later.

Full hidden costs breakdown →

Intelligence sourced from 2 independent sources
grounded summary Hacker News Tech community
Key claims include inline source attribution. Data verified against multiple independent sources. 7 source citations total.

NVIDIA NIM Contract Terms

NVIDIA NIM contracts do not auto-renew. Changes require advance notice. These terms are sourced from verified buyer experiences.

Contract Terms
Minimum Commitment 1 to 5 years for subscriptions, or one year for support service renewals
Based on 1 verified source

How to Negotiate NVIDIA NIM Pricing

NVIDIA NIM contracts are negotiable. These 1 tactics are sourced from real buyer experiences and procurement specialists.

Negotiation Playbook 1 tactics
Multi-year Discounts high success

NVIDIA offers multi-year discounts on AI Enterprise licenses, for example, a 5-year term for the perpetual license effectively reduces the annual cost per GPU.

https://vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIYQHIZZaAH34bvnoHhs5m1AdkM1IuvwbqeTOkUZNNyfKj3bq1dpbOyQ0n3BIP4iPzsEmpC-8d5ynXBtTM-nq-MGwYt8ROihze3UKtoNede3cRX8g1KQTNrfhnh675nmXBaNdtpY-9U5KiYm5zMYM68odYKj8W6qBEdrgwt_Bwbk1YHnL6uRhiDl23RmHtonE=
High success Medium Low / difficult

Full negotiation guide →

NVIDIA NIM Pricing FAQ

01 Does NVIDIA NIM have a free plan?

Yes. NVIDIA NIM includes a Developer (Free credits) tier at $0/month, giving developers access to hosted NIM endpoints for prototyping. When free credits are exhausted, the Pay-as-you-go (hosted NIM endpoints) tier applies based on token consumption.

02 How much does NVIDIA NIM cost per token on hosted endpoints?

Via OpenRouter, NVIDIA NIM-powered models range from $0.04 to $1.20 per million input tokens and $0.16 to $1.20 per million output tokens depending on the model. The provider median across 6 models is $0.095/1M input and $0.425/1M output. The largest model, Llama 3.1 Nemotron 70B Instruct, is $1.20/1M for both input and output tokens. The smallest model, Nemotron Nano 9B V2, starts at $0.04/1M input. Note: OpenRouter pricing reflects router-level rates and may include a modest markup over direct NVIDIA API pricing.

03 What does the NVIDIA NIM Enterprise plan include?

The Enterprise plan (AI Enterprise license + DGX Cloud) is custom-priced and intended for organizations that need to deploy NIM microservices at scale, either on NVIDIA DGX Cloud or on-premises infrastructure with enterprise support and SLAs. Contact NVIDIA sales for pricing.

04 Which models are available via NVIDIA NIM hosted endpoints?

Models available via hosted NIM endpoints (as listed on OpenRouter) include: Llama 3.1 Nemotron 70B Instruct ($1.20/1M input, $1.20/1M output), Nemotron Nano 12B V2 VL ($0.20/1M input, $0.60/1M output), Llama 3.3 Nemotron Super 49B V1.5 ($0.10/1M input, $0.40/1M output), Nemotron 3 Super 120B ($0.09/1M input, $0.45/1M output), Nemotron 3 Nano 30B ($0.05/1M input, $0.20/1M output), and Nemotron Nano 9B V2 ($0.04/1M input, $0.16/1M output).

05 Is NVIDIA NIM free to use?

Yes, NVIDIA NIM offers a Developer (Free credits) plan that provides access to hosted NIM endpoints and developer API access to Nemotron and Llama-based NIM models. However, this is limited to a free credits allocation and is not designed for production workloads. Once free credits are exhausted, you will need to move to pay-as-you-go or an Enterprise plan.

06 How much does NVIDIA NIM cost for pay-as-you-go usage?

Pay-as-you-go pricing for hosted NIM endpoints varies by model. Based on OpenRouter pricing data for NVIDIA models, rates range from $0.05/1M input tokens (Nemotron 3 Nano 30B) up to $0.50/1M input tokens and $2.20/1M output tokens (Nemotron 3 Ultra 550B). The provider median across 4 sampled models is approximately $0.24/1M input tokens and $0.425/1M output tokens.

07 When should I upgrade from the free Developer plan?

You should upgrade from the Developer (Free credits) plan when your free developer credits are exhausted, when you need production-scale throughput or reliability guarantees, or when you require on-premises or DGX Cloud enterprise deployment.

08 Is there a free tier for NVIDIA NIM?

Yes. The Developer (Free credits) tier provides access to hosted NIM endpoints and developer API access to Nemotron and Llama-based NIM models. However, this is not unlimited perpetual usage — it is limited to a free credits allocation and is not designed for production workloads. Once your free developer credits are exhausted, you will need to move to the Pay-as-you-go or Enterprise tier.

09 Can I self-host NVIDIA NIM microservices on the free tier?

No. The Developer (Free credits) tier gives access to hosted NIM endpoints only. Self-hosting NIM microservices on your own infrastructure requires the Enterprise tier (NVIDIA AI Enterprise license).

Is this pricing incorrect? — we'll verify and update it.

1 AI answer about this product’s pricing was flagged for review against CostBench’s record. Claim this record free to see them