托管云Nova 2 Lite
数据驻留：未公开 · 隐私：未公开 · SLA：未公开来源：Amazon Bedrock regional availability by models ↗选择要核对的信息，直接进入对应章节；各项事实均保留版本、条件和来源。
10/14 项有公开值 · 当前版本 Nova 2 Lite · 数据截止 2026-10-20
页面只展示已审核 Claim；待审核数据不会被当作公开事实。只展示已审核事实；未公开、冲突和过期状态不会被补猜。
只判断选型资料是否足够，不评价模型能力高低。
所有必填事实均通过状态、来源、证据与时效校验。最早核验时间：2026-09-17。
release_datelifecycle_statuscapability_tagsdeployment_optionscontext_window_tokensapi_idinput_price_usd_per_mtokoutput_price_usd_per_mtokavailabilityNova 2 Lite 是 Amazon AI 的 Amazon Nova 2 模型，年鉴当前记录 1 个版本，最新公开记录为 Nova 2 Lite。
权重为未披露，代码为未披露，训练数据为未披露，部署方式为managed_api；这些维度分别判断，不能合并成单一“开源/闭源”结论。
当前公开快照的上下文窗口为 1M tokens，架构为 未公开；这里只使用通过审核且来源允许公开采集的 Claim。
5 个维度暂无可公开投影的审核记录
补充来源或线索会进入投稿队列，经编辑核验后才会成为公开事实。当前来源记录 3 个。
技术事实、业务价值、企业治理和证据质量分区呈现
厂商、家族、API ID、日期与生命周期
Dense/MoE、参数、层数、专家与 Tokenizer
预训练、SFT、偏好优化与安全对齐
教师/学生、基础模型、轨迹和限制
权重、代码、数据、许可证、部署与 API
API、托管云、地区、集成、隐私与 SLA
输入输出、缓存、Batch 与部署成本边界
数据、权限、审计、合规、安全与人工兜底
12 条已审核事实 · 全部来自公开来源闸门通过的记录
尚无已审核公开记录的维度：02 能力与任务、09 性能与评测、11 业务场景、12 落地与提效。查看资料缺口 ↑
关系只来自年鉴数据库中已登记的归属字段。
接口兼容与业务适配均以来源绑定记录为准，不由模型名称或能力标签推断。
“兼容接口/平台”仅代表来源明确登记的接入边界，不等同于全部第三方工具都可直接使用。
下载、API、客户端和更新内容只展示已登记的来源事实；未登记不作推断。
not_disclosed
获取入口：Nova 2 Lite model card ↗26 个已登记地区 · 43 条版本级记录
The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
保留发布日期精度、生命周期节点和原始来源，未公开日期不会被推断。
模型版本 · active
显示最新核验的 8 条 · 共 12 条已审核事实
The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
1 个版本 · 每项可回到来源与技术解释
高置信度，已绑定来源与证据。
条件：Programmatic Access table on the model card lists amazon.nova-2-lite-v1:0 for the bedrock-runtime Invoke and Converse APIs.
来源：Nova 2 Lite model card ↗The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
高置信度，已绑定来源与证据。
条件：Derived only from the card: Image, Text and Video inputs make the model a video and image understanding model; client-side tool calling is listed as supported, which is the basis for the agent tag.
来源：Nova 2 Lite model card ↗引用文本与接口返回都只包含已审核公开事实
Amazon AI Nova 2 Lite｜最新版本 Nova 2 Lite｜已审核事实 12 条｜公开来源 3 个｜数据截止 2026-10-20｜AI 年鉴｜https://ainianjian.com/models/model.amazon.nova-2-lite
公开接口返回该模型的版本、已审核事实 12 条及其来源信息，可直接用于脚本或对比。
https://api.ainianjian.com/api/models/model.amazon.nova-2-lite ↗curl -s 'https://api.ainianjian.com/api/models/model.amazon.nova-2-lite'
技术规格使用当前版本已审核且来源允许公开采集的事实；模型级开放信息与跨版本地区记录分别标注范围。请按每项适用范围核对，未登记字段显示「未公开」，不使用估算值填充。
Amazon publishes a model-by-model matrix with Region rows and three routing columns: In-Region, Geographic Cross-Region, and Global Cross-Region. This curated slice stores only combinations explicitly marked available for the six existing Amazon Nova versions; it does not infer support for unlisted combinations and does not treat the Bedrock service footprint or managed API access as model-specific regional support.
Amazon publishes a model-by-model matrix with Region rows and three routing columns: In-Region, Geographic Cross-Region, and Global Cross-Region. This curated slice stores only combinations explicitly marked available for the six existing Amazon Nova versions; it does not infer support for unlisted combinations and does not treat the Bedrock service footprint or managed API access as model-specific regional support.
The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
The AWS Bedrock pricing data for us-east-1 lists Amazon Nova 2 Lite on-demand Global cross-region inference at 0.0003 USD per 1K input tokens and 0.0025 USD per 1K output tokens, equivalent to 0.30 and 2.50 USD per million tokens. In-Region token rates are published but the model card reports no In-Region availability for this model, so the cross-region Global rate is the applicable list price.
The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
高置信度，已绑定来源与证据。
条件：Re-verified on 2026-10-20 against the current model card; the card reports Context window: 1M tokens.
来源：Nova 2 Lite model card ↗The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
高置信度，已绑定来源与证据。
来源：What is Amazon Nova 2? ↗Amazon states that Nova 2 includes Nova 2 Lite, Nova 2 Sonic, and Nova Multimodal Embeddings; the documentation describes up to 1 million tokens of context and up to 65,536 generated tokens for the Nova 2 model documentation.
高置信度，已绑定来源与证据。
条件：On-demand Global cross-region inference, USD 0.0003 per 1K input tokens. In-Region is not offered for this model.
来源：Amazon Bedrock pricing ↗The AWS Bedrock pricing data for us-east-1 lists Amazon Nova 2 Lite on-demand Global cross-region inference at 0.0003 USD per 1K input tokens and 0.0025 USD per 1K output tokens, equivalent to 0.30 and 2.50 USD per million tokens. In-Region token rates are published but the model card reports no In-Region availability for this model, so the cross-region Global rate is the applicable list price.
高置信度，已绑定来源与证据。
条件：Re-verified on 2026-10-20; the card reports Knowledge cutoff: Oct 2025, so the value keeps month precision.
来源：Nova 2 Lite model card ↗The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
高置信度，已绑定来源与证据。
条件：Model lifecycle: Active on 2026-10-20, with EOL no sooner than Dec 02, 2026.
来源：Nova 2 Lite model card ↗The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
高置信度，已绑定来源与证据。
条件：Re-verified on 2026-10-20; the card reports Max output tokens: 64K.
来源：Nova 2 Lite model card ↗The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
高置信度，已绑定来源与证据。
条件：Re-verified on 2026-10-20; the card links the Amazon Bedrock End User License Agreements and Terms of Use and publishes no separate open-weight license.
来源：Nova 2 Lite model card ↗The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
高置信度，已绑定来源与证据。
条件：On-demand Global cross-region inference, USD 0.0025 per 1K output tokens. Batch and Priority tiers have separate rates.
来源：Amazon Bedrock pricing ↗The AWS Bedrock pricing data for us-east-1 lists Amazon Nova 2 Lite on-demand Global cross-region inference at 0.0003 USD per 1K input tokens and 0.0025 USD per 1K output tokens, equivalent to 0.30 and 2.50 USD per million tokens. In-Region token rates are published but the model card reports no In-Region availability for this model, so the cross-region Global rate is the applicable list price.
高置信度，已绑定来源与证据。
条件：Re-verified on 2026-10-20; the card reports Model launch date: Dec 02, 2025.
来源：Nova 2 Lite model card ↗The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.
高置信度，已绑定来源与证据。
条件：Re-verified on 2026-10-20; the card publishes no weights download or repository. Do not infer closed weights from this absence.
来源：Nova 2 Lite model card ↗The Amazon Bedrock model card for Amazon Nova 2 Lite states: Model lifecycle: Active; Context window: 1M tokens; Max output tokens: 64K; Knowledge cutoff: Oct 2025; Model launch date: Dec 02, 2025; Model EOL no sooner than Dec 02, 2026; Input modalities: Image, Text, Video; Output modalities: Text; supported APIs: Invoke and Converse on bedrock-runtime; Model ID: amazon.nova-2-lite-v1:0. The Regional availability table marks In-Region as unavailable for every listed region while Geo and Global cross-region inference are available. The features table lists response streaming, abuse detection, guardrails, prompt optimization and client-side tool calling as supported, and intelligent prompt routing, count tokens, knowledge base and structured outputs as not supported.