Claude Fable 5$22.000/MClaude Opus 4.8$11.000/MClaude Opus 4.7$11.000/MClaude Opus 4.6$11.000/MClaude Opus 4.5$33.000/MClaude Sonnet 3.7$6.600/MClaude Opus 3$33.000/MClaude 2.1$12.800/MClaude 2$12.800/MGPT-5.5$12.500/MGPT-5.2$5.425/MGPT-5.2-Codex$5.425/MGPT-5$3.875/MGPT-4.5$97.500/MGPT-4 Turbo Preview$16.000/MGPT-4$39.000/MGPT-4-32k$78.000/Mo3$19.000/Mo3-mini$2.090/Mo4-mini$2.090/Mo1$28.500/Mo1-mini$5.700/Mo1-preview$28.500/MGemini 3.5 Pro$5.000/MGemini 3.1 Pro$5.000/MGemini 3 Pro$5.000/MGemini 2.5 Pro$3.875/MGemini 1.5 Pro$2.375/MGemini 1.0 Ultra$12.000/MGemini 1.0 Pro$0.800/MClaude Fable 5$22.000/MClaude Opus 4.8$11.000/MClaude Opus 4.7$11.000/MClaude Opus 4.6$11.000/MClaude Opus 4.5$33.000/MClaude Sonnet 3.7$6.600/MClaude Opus 3$33.000/MClaude 2.1$12.800/MClaude 2$12.800/MGPT-5.5$12.500/MGPT-5.2$5.425/MGPT-5.2-Codex$5.425/MGPT-5$3.875/MGPT-4.5$97.500/MGPT-4 Turbo Preview$16.000/MGPT-4$39.000/MGPT-4-32k$78.000/Mo3$19.000/Mo3-mini$2.090/Mo4-mini$2.090/Mo1$28.500/Mo1-mini$5.700/Mo1-preview$28.500/MGemini 3.5 Pro$5.000/MGemini 3.1 Pro$5.000/MGemini 3 Pro$5.000/MGemini 2.5 Pro$3.875/MGemini 1.5 Pro$2.375/MGemini 1.0 Ultra$12.000/MGemini 1.0 Pro$0.800/M
BETA
Subscribe
← Back to Index·All Nvidia models
Nvidia
NvidiaEfficient

Phi-3-Mini-4K (NIM)

Nano

Microsoft Phi-3 Mini hosted as an NVIDIA NIM. Same Phi-3 weights, packaged for NVIDIA AI Foundry deployment.

Phi-3-Mini-4K (NIM) is a efficient AI model from Nvidia. It costs $0.040 per million input tokens and $0.040 per million output tokens (blended $0.040/M), with a 4K-token context window.

Modalities textOfficial model page ↗API docs ↗Compare with another model →Estimate monthly cost →
INPUT
$0.040/M
per million input tokens
OUTPUT
$0.040/M
per million output tokens
BLENDED 70/30
$0.040/M
default reference rate · how it's calculated →
CONTEXT
4K
4,000 tokens
What it's good at
  • Phi-3 mini quality
  • NIM packaging
  • On-device or NVIDIA-stack deployment
Typical use cases
  • NVIDIA-stack edge inference
  • NIM-based small-model deployments
Benchmarks
vs. best public score
Scores inherited from Phi-3 Mini — this is a hosted variant of the same open-weights model, so the underlying benchmark scores are identical.
MMLU69%
Multitask academic knowledge across 57 subjects.
MATH49%
High-school competition math problems.
HumanEval59%
Python function synthesis from docstrings.
Hand-curated from each provider's published reports and public leaderboards. Methodology varies across sources — treat as directional rather than authoritative.
More from Nvidia
See all 6 →
Llama-3.1-Nemotron-70B
RLHF-Aligned · 128K ctx
in $0.350/Mout $0.400/M
Llama-3.1-Nemotron-Ultra-253B
Ultra · 128K ctx
in $1.600/Mout $1.600/M
Nemotron-4-340B
Large · 4K ctx
in $4.200/Mout $4.200/M
Mistral-NeMo-12B (NIM)
Efficient · 128K ctx
in $0.150/Mout $0.150/M
Mistral-Large-2 (NIM)
Enterprise · 128K ctx
in $2.000/Mout $6.000/M
Frequently asked questions

How much does Phi-3-Mini-4K (NIM) cost?

Phi-3-Mini-4K (NIM) costs $0.040 per million input tokens and $0.040 per million output tokens, for a blended reference rate of $0.040 per million tokens.

What is Phi-3-Mini-4K (NIM)'s context window?

Phi-3-Mini-4K (NIM) supports up to 4K tokens of context (4,000 tokens).

What is Phi-3-Mini-4K (NIM) best for?

Phi-3-Mini-4K (NIM) is well suited to Phi-3 mini quality, NIM packaging and On-device or NVIDIA-stack deployment.

Who makes Phi-3-Mini-4K (NIM)?

Phi-3-Mini-4K (NIM) is developed and served by Nvidia.