Model

Amazon Nova Premier

CLOSED

Amazon · Amazon Nova family · released Apr 30, 2025

Amazon's most capable Nova: 1M-token context for complex agentic/long-document tasks; explicitly not a reasoning model, positioned mainly as a distillation teacher for smaller Nova models.

ReasoningCodingVisionFunction callingTool useAgentic

1235.4

Elo · rank #175

Parameters

Undisclosed

Active params

—

Context

1M tokens

Architecture

Transformer, architecture/size undisclosed

License

Proprietary (AWS Bedrock service terms)

Languages

200+

API price (in/out)

$2.5 / $12.5

Modalities

text · vision · video

Benchmark results

Bar shows position within the tracked field; marker = field best

AIMEMath16.0%#143

best: GPT-5.2 · 100.0%

BFCL v3Agents63.7%#24

best: Hunyuan-A13B · 78.3%

BigCodeBenchCoding28.1%#8

best: GPT-4o mini · 57.4%

CharXivVision48.8%#31

best: Muse Spark 1.1 (xhigh) · 88.4%

GPQA DiamondReasoning56.9%#170

best: GPT-5.6 · 94.6%

IFEvalReasoning91.5%#11

best: Gemma 4 26B A4B · 98.5%

MATH-500Math82.0%#64

best: GPT-5 · 99.4%

MMLU-ProKnowledge73.3%#82

best: Claude Fable 5 · 91.5%

MMLUKnowledge87.4%#31

best: OpenAI o3 · 92.9%

MMMUVision68.0%#62

best: Claude Fable 5 · 89.3%

MT-BenchHuman preference8.6#22

best: Hunyuan-Large (A52B) · 9.4

SWE-bench VerifiedCoding42.4%#99

best: Claude Fable 5 · 95.0%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

$2.5

Output / M tok

$12.5

Amazon Nova family

Elo progression across releases

Amazon Nova LiteDec 20241039.0 Amazon Nova MicroDec 2024923.2 Amazon Nova ProDec 20241216.3 Amazon Nova PremierApr 20251235.4

Resources

DOCSdocs.aws.amazon.com/bedrock/latest/userguide/model-card-amazon-nova-premier.html APIAvailable via Amazon API

API price $2.5/$12.5 · each benchmark row carries its own source badge (see methodology)