Benchmark Almanac

Amazon Nova Premier

Amazon Nova Premier.

Amazon Released April 30th, 2025 Family Amazon Nova The lab's model card Compare with others
27
Benchmarks with a score
0
Ranked first, out of five or more models
0
In the top three
Amazon Apr 2025

Amazon Nova Premier

Closed weights Flagship

126.5 #306 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • GPQA Main 51.8% #34 /60
  • SnorkelSequences 51.8% #14 /25
  • SnorkelGraph 34.5% #15 /24

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
SimpleQA 86.3% #6 of 60 Lab model card
IFEval 91.5% #23 of 135 Lab model card
MMLU 87.4% #35 of 131 Lab model card
SnorkelSpatial 14.2% #22 of 27 Official leaderboard

Math and science

BenchmarkScoreRankSource
SnorkelSequences 51.8% #14 of 25 Official leaderboard
GPQA Main 51.8% #34 of 60 Official leaderboard
SnorkelGraph 34.5% #15 of 24 Official leaderboard
GPQA Diamond 57.1% #229 of 302 Lab model card
MATH-500 82% #60 of 70 Lab model card
AIME 2025 16% #141 of 143 Lab model card

Coding

BenchmarkScoreRankSource
MBXP 78.4% #1 of 3 Lab model card
SWE-bench Verified 42.4% #38 of 60 Official leaderboard
BigCodeBench-Hard 28.1% #2 of 3 Lab model card
DesignArena Website 828 #146 of 151 Official leaderboard

Agents and tools

BenchmarkScoreRankSource
ScreenSpot 87.8% #6 of 11 Lab model card
Berkeley Function Calling Leaderboard (BFCL) 63.7% #34 of 41 Lab model card

Professional work

BenchmarkScoreRankSource
SnorkelUnderwrite 57% #15 of 27 Official leaderboard
FinQA (Snorkel OpenEnv environment) 62.1% #17 of 22 Official leaderboard
SnorkelFinance 62.1% #17 of 22 Official leaderboard
Finance Reasoning (Snorkel) 34.2% #17 of 21 Official leaderboard

Vision, audio and robotics

BenchmarkScoreRankSource
TallyQA 61.5% #1 of 4 Lab model card
MMMU 68% #40 of 136 Official leaderboard
EgoSchema 73.8% #4 of 12 Lab model card
OCRBench v2 56.9% #18 of 22 Lab model card
CharXiv 48.8% #71 of 75 Lab model card
CharXiv (DQ) 84.6% #7 of 7 Lab model card

Other

BenchmarkScoreRankSource
HELM Capabilities 63.7 #35 of 60 Official leaderboard

Its lab's numbers put it at 126.5, number 306 of 404

The Self-Reported Index fits one score to the 13 capability results Amazon reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 120.4 to 132.6. How we compute it.

Its lab's numbers next to 3 leaderboards

Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.

Benchmark Lab says Board says Gap
IFEval 91.5% IFEval 80.3% Leaderboard, strict, share of instructions followed +11 pts
MMMU 68% MMMU 68% Leaderboard Same
SWE-bench Verified 42.4% SWE-Bench Verified (internal agentic scaffold) 42.4% Leaderboard, Amazon Nova Premier 1.0 Same

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Amazon Nova Premier: Our most capable model for complex tasks and teacher for model distillation (AWS News Blog) Apr 30, 2025

Its system card

Every result in the longer documents the lab published with it. 1 of them isn't in the launch post.