Nemotron Nano 12B V2

NVIDIA128K context

Open weightsVisionTool calling
Lowest · Fireworks AI
$0.20 / $0.20
per 1M in / out

As of Nov 18, 2026, Nemotron Nano 12B V2 costs $0.20 per million input tokens and $0.20 per million output tokens from Fireworks AI. The cheapest of 3 providers is Fireworks AI at $0.20 / $0.20. No price change since Appndo started tracking it on Oct 14, 2026. Recent price changes

Providers (3)

Sorted by your monthly estimate. Prices per million tokens. Fireworks AI also lists up to 131K output tokens per request.

  • Fireworks AICheapest

    Input / output
    $0.20 / $0.20
    Context
    128K
    Per month
    $2.40

    fireworks_ai/accounts/fireworks/models/nvidia-nemotron-nano-12b-v2

  • Input / output
    $0.20 / $0.60
    Context
    128K
    Per month
    $3.20

    nvidia.nemotron-nano-12b-v2

  • Input / output
    $0.20 / $0.60
    Context
    128K
    Per month
    $3.20

    bedrock_mantle/nvidia.nemotron-nano-12b-v2

Cloud and hosted providers may add regional pricing, minimums or platform fees. Supports: Vision, Tool calling.

Similar models

Common questions

How much does the Nemotron Nano 12B V2 API cost?

As of Nov 18, 2026, Nemotron Nano 12B V2 costs $0.20 per million input tokens and $0.20 per million output tokens from Fireworks AI.

Where is Nemotron Nano 12B V2 cheapest?

As of Nov 18, 2026, the cheapest of 3 providers is Fireworks AI at $0.20 per million input tokens and $0.20 per million output tokens. Appndo lists no official price from NVIDIA.

What is the context length of Nemotron Nano 12B V2?

Nemotron Nano 12B V2 takes up to 128K tokens of context at Fireworks AI.

How does Nemotron Nano 12B V2 compare with similar models?

Nemotron Nano 12B V2 has no LMArena score yet. Other NVIDIA models and their cheapest price per million input / output tokens: Nvidia Nemotron Nano 9B V2 ($0.04 / $0.16), Nvidia Nemotron 3 Ultra 550B A55B ($0.50 / $2.15), and Nemotron 3 Ultra 550B A55B ($0.80 / $2.60).

About this data

Prices come from LiteLLM's model cost map, with more providers from models.dev (both MIT licence), checked Nov 18, 2026; confirm on the provider's pricing page before you rely on a number.