Amazon model product
A low-cost multimodal model that is lightning fast for processing images, video, documents, and text.
Updated Aug 12, 2026. Default version: Nova Lite
Technical details for the model's default version.
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Benchmark scores for Nova Lite.
| VATEX | 77.8% | 01 | 2 | 100.0% | C | |
| CRAG | 43.8% | 02 | 3 | 50.0% | C | |
| FinQA | 73.6% | 02 | 3 | 50.0% | C | |
| GroundUI-1K | 80.2% | 02 | 2 | 0.0% | C | |
| MM-Mind2Web | 60.7% | 02 | 3 | 50.0% | C | |
| Translation en→Set1 COMET22 | 88.8% | 02 | 3 | 50.0% | C | |
| Translation en→Set1 spBleu | 41.5% | 02 | 3 | 50.0% | C | |
| Translation Set1→en COMET22 | 88.8% | 02 | 3 | 50.0% | C | |
| Translation Set1→en spBleu | 43.1% | 02 | 3 | 50.0% | C | |
| VisualWebBench | 77.7% | 02 | 2 | 0.0% | C | |
| SQuALITY | 19.2% | 04 | 5 | 25.0% | C | |
| BBH | 82.4% | 06 | 12 | 54.5% | C | |
| EgoSchema | 71.4% | 06 | 9 | 37.5% | C | |
| ARC-C | 92.4% | 08 | 34 | 78.8% | C | |
| TextVQA | 80.2% | 08 | 15 | 50.0% | C | |
| BFCL | 66.6% | 09 | 11 | 20.0% | C | |
| ChartQA | 86.8% | 10 | 24 | 60.9% | C | |
| DROP | 80.2% | 12 | 30 | 62.1% | C | |
| GSM8k | 94.5% | 15 | 48 | 70.2% | C | |
| DocVQA | 92.4% | 16 | 26 | 40.0% | C | |
| IFEval | 89.7% | 19 | 65 | 71.9% | C | |
| LVBench | 40.4% | 24 | 24 | 0.0% | C | |
| MATH | 73.3% | 31 | 71 | 57.1% | C | |
| HumanEval | 85.4% | 35 | 66 | 47.7% | C | |
| MMMU | 56.2% | 50 | 63 | 21.0% | C | |
| MMLU | 80.5% | 59 | 100 | 41.4% | C | |
| GPQA | 42.0% | 200 | 234 | 14.6% | C |
Preference and agent-evaluation results for the default version.
| vision | overall | 130 | 990.6 | 1,854 | N/A | |
| vision style control | overall | 131 | 1019.3 | 1,854 | N/A | |
| text | overall | 273 | 1228.6 | 19,372 | N/A | |
| text style control | overall | 288 | 1260.4 | 19,372 | N/A |
Provider-specific output speed and catalog latency for Nova Lite. Runtime does not affect the capability score.
| Bedrock | 100 tok/s | 0.5 s | 300K | 2K |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Official vendor API pricing appears first, followed by individual provider offers.
| Amazon Bedrock | amazon.nova-lite-v1:0 | global | $0.06 | $0.24 | 300K | |
| Vercel AI Gateway | amazon/nova-lite | global | $0.06 | $0.24 | 300K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Available versions of this model. The score column identifies the version used in the overall ranking.
| Nova Lite | 13.2 | N/A | 300K | 2K | No | Proprietary |
Key information about Nova Lite and its available data.
A low-cost multimodal model that is lightning fast for processing images, video, documents, and text.
Data as of 2026-08-11.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
Common questions about Nova Lite.
Nova Lite's default version was released on Nov 20, 2024.
Nova Lite's official API price is $0.06 per million input tokens and $0.24 per million output tokens via Amazon Bedrock. The lowest tracked third-party offer starts at $0.06 input and $0.24 output via Vercel AI Gateway.
Nova Lite was created by Amazon.
The default version has a 300K token context window.
No. The default version is not marked as having publicly available weights.
2 provider offerings are linked to the default version.
Nearby ranked alternatives include GPT-4-Turbo, Llama 4 Scout, Phi 4 Mini Reasoning.