Amazon
Nova 2 Lite
Release Date 2025-12-021M IN · 66K OUTProprietary ModelAPI available
Amazon Nova 2 Lite is a fast, cost-effective reasoning model released by AWS, capable of processing text, images, and video inputs. It features a 1M-token context window, configurable thinking effort (low/medium/high), and built-in tools such as code interpreter and web grounding, making it well-suited for everyday enterprise workloads like document processing, information extraction, and multi-step agentic workflows. Nova 2 Lite delivers competitive benchmark performance against models such as Claude Haiku 4.5 and Gemini 2.5 Flash, while maintaining significantly lower cost.
API cost
Provider price
$0.3IN$2.5OUT
SourceOpenRouterLast updated
Output Speed
Very fastper 2,000 tokens
113tok/s
First Output 0.5stotal: 18.2s
SourceOpenRouterLast updated
User Comparisons
Writing Style
Answer length
Shorter side
top 66%
Vocabulary diversity
Plainer side
top 73%
Sentence rhythm
Average
top 55%
Language mixing
Average
top 50%
List usage
Very often
48% of lines
Heading usage
Occasionally
14% of lines
Reasoning share
Minimal
0% of total
Lies (Hallucination)
Honest answers
50%
Correct 0Honest don't-know 3Lies 3
Lies without doubt
1 questions
Response Speed
instability ±402%
First response 1.5sTotal 8.6s
136 tok/s
6.1ms per token
User Reviews
- •
Quite fast and accurate answers. Sometimes shows ability beyond expectations.
- •
Answers are super fast and the content is accurate. Excellent performance for a lightweight model. Korean is also natural.
SourceAIB Inc.Last updated
Expert Performance Evaluation
Overall performance
All overall performance AI rankings13.4Reasoning effort: High133581.8%77.0%
By effortNone 8.70Low 11.8Medium 12.5High 13.4
Automation (agents)
All automation AI rankingsCan it handle customer service by the rules?TAU2Bottom 32%#55
75.7%Reasoning effort: Medium
By effortNone 62.0%Low 71.9%Medium 75.7%High 72.8%
Documents and office work
All documents and office work AI rankings60.3%Reasoning effort: High1351
By effortNone 18.7%Low 54.3%Medium 60.0%High 60.3%
Research and fact-checking
All research and fact-checking AI rankingsWhich AI gives answers people prefer most, without false claims?Arena Text FactualityBottom 5%#87
1380
Math and science
All math and science AI rankings11.6%Reasoning effort: High133281.1%94.394.3%13571347
By effortNone 2.9%Low 4.0%Medium 9.0%High 11.6%
Image understanding
All image understanding and analysis AI rankingsCan it solve college-level problems with images and diagrams?MMMU-ProBottom 6%#36
61.8%Self-reported by developer
See each evaluation’s detail page for its source and last updateData sources & removal requests
Related Info
Public
Amazon🇺🇸 Seattle1994$2.8T~1.55M