Compare AI models

Specifications
Parameters
4B—
Context window
256k400k
API pricing
Input price
—$0.25
Output price
—$2.00
Cached input price
—$0.025
Cheapest input
—$0.125OpenAI
Cheapest output
—$1.00OpenAI
—$150OpenAI list price
Benchmarks
τ²-bench · retail
39.8%78.3%
τ²-bench · airline
33.3%60%
τ²-bench · telecom
33%74.1%
AIME 2025
78.5%91.1%
GPQA Diamond
53.2%82.3%
Overview
CompanyNVIDIAOpenAI
Release dateMar 16 2026Aug 7 2025
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

GPT-5 mini leads Nemotron 3 Nano 4B on 5 of the 5 benchmarks they both report (τ²-bench, AIME 2025, GPQA Diamond). Only GPT-5 mini has a verified first-party price for an available API: $0.25 per million input tokens and $2.00 per million output tokens. No currently available pay-as-you-go API rate is tracked for Nemotron 3 Nano 4B. GPT-5 mini shipped 221 days before Nemotron 3 Nano 4B, so benchmark comparisons should account for the intervening progress.

Context windows are 256k (Nemotron 3 Nano 4B) vs 400k (GPT-5 mini). Nemotron 3 Nano 4B is open weight, while GPT-5 mini is closed.

On τ²-bench · retail, GPT-5 mini leads at 78.3% vs Nemotron 3 Nano 4B at 39.8%. On τ²-bench · airline, GPT-5 mini leads at 60% vs Nemotron 3 Nano 4B at 33.3%. On τ²-bench · telecom, GPT-5 mini leads at 74.1% vs Nemotron 3 Nano 4B at 33%. On AIME 2025, GPT-5 mini leads at 91.1% vs Nemotron 3 Nano 4B at 78.5%. On GPQA Diamond, GPT-5 mini leads at 82.3% vs Nemotron 3 Nano 4B at 53.2%.