Benchmark Almanac

Amazon

8 models 5 model cards read amazon.science

What its model cards measure

Tall ticks are Amazon model cards that list the benchmark in their headline table. Gray dots are cards that leave it out. Rows are the benchmarks the lab listed most, at least twice each, in the order it started using them.

ScreenSpot AIME 2025 OCRBench v2 GPQA Diamond SWE-bench Verified MMLU-Pro IFBench MultiChallenge tau2-bench tau2-bench Verified BFCL V4 MMMU-Pro RealKIE-FCC Verified QVHighlights LongCodeBench Common Voice Terminal-Bench LiveCodeBench v5

Model cards

DateModelDocumentBenchmarks
2 Dec 2025 Nova 2 Omni Preview Amazon Nova 2: Multimodal Reasoning and Generation Models (Technical Report and Model Card) 19
2 Dec 2025 Nova 2 Lite Amazon Nova 2: Multimodal Reasoning and Generation Models (Technical Report and Model Card) 18
2 Dec 2025 Nova 2.0 Pro Preview Amazon Nova 2: Multimodal Reasoning and Generation Models (Technical Report and Model Card) 18
30 Apr 2025 Amazon Nova Premier Amazon Nova Premier: Our most capable model for complex tasks and teacher for model distillation (AWS News Blog) 16
31 Mar 2025 Amazon Nova Act Introducing Amazon Nova Act (Amazon Science) 2

Every Amazon model