视频版Nano Banana来了:内置Gemini世界知识,原版香蕉出图仅需4秒

marsbitPublicado em 2026-07-01Última atualização em 2026-07-01

Resumo

谷歌发布两款新型多模态AI模型:视频生成模型Gemini Omni Flash与高速图像生成模型Nano Banana 2 Lite。 Gemini Omni Flash融合了Gemini的世界知识与视频生成能力,支持通过文本、图像或视频输入生成或编辑高质量视频,具备对话式编辑、多模态参考等核心功能,生成成本约为每秒0.10美元。不过目前仅支持生成10秒视频,且人物一致性等方面仍有局限。 Nano Banana 2 Lite则主打高速与经济性,生成一张1K分辨率图像仅需约4秒、成本约0.034美元,速度达前代产品的5倍,且文字渲染效果出色。 谷歌特别展示了将两款模型串联使用的应用潜力:可先用Nano Banana 2 Lite快速生成图像,再交由Omni Flash转化为动态视频,并演示了“赛博旅游”、“虚拟装修预览”和“电商短视频自动生成”三个具体场景,体现了多模态AI在垂直领域的落地价值。尽管在代码能力上不占优势,但谷歌正凭借其多模态技术布局,深入电商、内容创作等实用赛道,探索差异化的AGI发展路径。

虽然Coding还是一坨,但谷歌搞「多模态」确实有两把刷子。

Gemini Omni Flash正式开放API,视频版Nano Banana

麻瓜爆改「哈利波特」不再是梦想。且看官方用Gemini Omni表演的这四个数字魔术:

太离谱了,这一致性和文字清晰度,还要啥绿幕和特效,直接现场直播奇异博士。

与此同时,大家心心念念的「香蕉」,也迎来了「光速版」。

Nano Banana 2 Lite:迄今为止最快、最经济高效的Gemini图像模型。

毫不夸张——4秒出一张图,1K分辨率图像仅需约2毛多

和Nano Banana 2放在一起对比,这个速度,简直起飞了。

更别说一次生图能干3分钟的GPT Image 2......

怪不得半天没憋出来Gemini 3.5 Pro,敢情时间全花在心心念念的多模态上了是吧,哈萨比斯!!

Gemini Omni Flash

Google I/O 2026大会上首次亮相的Gemini Omni Flash,把Gemini的多模态推理能力与视频生成、编辑深度结合,当时就引发了不少关注。

如今,这个模型已通过Gemini API和Google AI Studio正式向开发者开放。能基于文本、图像和视频等多种输入,轻松生成高质量视频并进行编辑。

四项关键能力:

对话式视频编辑:用自然语言修改和精修视频,跟改飞书文档似的。

多模态参考:组合图像、文本、视频输入,保持场景控制和一致性。

现实世界知识:调用Gemini在历史、生物、叙事逻辑等方面的知识来构建视频,不用你再写三页Prompt描述建筑风格。

文字与动作同步:通过简单提示词,将文字和图形直接连接到视频动作。

价格也很有竞争力:每秒视频输出成本0.10美元,跟Veo 3.1 Fast持平。

定位方面,同为轻量版视频生成模型,Omni Flash更强调Gemini世界知识,生态方面也是全面朝Gemini层靠拢的。

不过谷歌也挺坦诚,主动列了一堆现阶段局限:

1、目前只支持10秒视频生成,后续会支持更长;

2、暂不支持音频参考上传和场景扩展;

3、API支持最长3秒的视频作为参考素材,但目前模型还无法正确处理这类输入;

4、场景切换和运镜时的人物一致性仍有局限。

Nano Banana 2 Lite

Nano Banana 2 Lite(又名gemini-3.1-flash-lite-image),专为高速处理而设计。

经过针对性优化,它瞄准的是那些对延迟极度敏感、需要短时间内批量处理大量图像的实时应用场景——比如电商素材批量生成、广告创意快速迭代、自动化内容流水线。

两大核心卖点——

光速:出图延迟约4秒,是Nano Banana 2的五分之一(后者约20秒)。

白菜价:一张1K图片仅需约0.034美元,是Nano Banana 2的一半,Nano Banana Pro的四分之一。

速度砍了,价格砍了,但生图和图像编辑能力并没有明显缩水。Nano Banana 2 Lite 依然保持了出色的文字渲染效果,在benchmark上和Grok等模型处于同一水平线。

因此谷歌的建议是:如果你还在贪便宜用初代Nano Banana,赶紧换。Lite版在各项关键指标上已经全方位碾压了。

双剑合璧

等等,先别走。

本以为这只是两款并行的模型发布,没想到谷歌表示:还有新花样。

真正的神奇之处,在于将这些模型串联起来使用。

众所周知,AIGC创作需要反复迭代,素材管理相当麻烦。

如今,借助这两个模型,终于不用再反复上传文件了,图像生成与视频创作无缝衔接。

具体来说,可以先用Nano Banana 2 Lite高速出图,再把生成的图像作为参考素材喂给 Gemini Omni Flash,一键转化为视频。

为了展示这套1+1>2的魔法Workflow,谷歌甚至专门做了3个Demo APP:

1、Anywhere

自拍一张或上传照片,NB2 Lite瞬间把你P到几十个地标景点里。

接着点一下图片,Omni Flash把静态景点变成动态短片。

赛博旅游,这下也端到端了。

2、Space Lift

这有点吓人,感觉再和Genie世界模型结合一下,未来怕是要威胁到不少传统装修方案SaaS公司。

上传一张房间照片,NB2 Lite先出各种装修风格方案。找到喜欢的,点视频按钮,Omni能直接给你一段电影级空间漫游。

3、Omni product studio

跨境电商的福音来了。

给产品拍张白底图,NB2 Lite出各种场景化商品图,Omni Flash再把静态图变成电商短视频。

从「产品」到「广告素材」,全链路自动跑完。

所以,多模态到底有什么用?

这个问题谷歌肯定被问了无数次。

尤其在2026年,Coding几乎等同于模型智商的代名词。每家都在Coding上往死里卷。

死磕多模态,图什么?

AGI叙事那一套就不说了,短期来看的话,谷歌这套多模态模型确实能赋能旗下不少产品。

AGI叙事那一套就不展开了。短期来看,谷歌这套多模态模型确实能赋能旗下不少产品——Stitch是一个,Pixel内置的P图是一个,NotebookLM的横空出世也蛮惊艳。

这次发的两个新模型,则让人看到了多模态在垂直场景落地的更多潜力。电商、装修、短视频......这些业务的需求是真的,钱也是真的。

加上有安卓生态加持,基本不用太担心商业化的问题。

Coding谷歌暂时追不上,但多模态这张牌桌,谷歌可能是唯一能组齐一套牌的玩家。

不过......

Gemni 3.5 Pro到底啥时候来啊!!!

参考链接:[1]https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-omni-flash-nano-banana-2-lite/

本文来自微信公众号“量子位”,作者:关注前沿科技

Criptomoedas em alta

Perguntas relacionadas

Q文章介绍的两款新模型分别叫什么名字,它们的核心定位是什么?

A文章介绍了Gemini Omni Flash和Nano Banana 2 Lite两款新模型。Gemini Omni Flash是一个轻量级视频生成模型,核心定位是将Gemini的多模态推理能力与视频生成、编辑深度结合,强调利用Gemini的世界知识。Nano Banana 2 Lite(又称gemini-3.1-flash-lite-image)则是一个图像生成模型,专为高速、低成本处理大量图像而设计,瞄准对延迟极度敏感的实时应用场景。

QNano Banana 2 Lite的图像生成速度和成本是多少?

ANano Banana 2 Lite出图延迟约为4秒,一张1K分辨率图像的生成成本约为0.034美元(约合人民币两毛多)。其速度是前代Nano Banana 2的五分之一(后者约20秒),价格是后者的一半。

Q根据文章,Gemini Omni Flash具备哪四项关键能力?

AGemini Omni Flash具备四项关键能力:1. 对话式视频编辑:用自然语言修改和精修视频。2. 多模态参考:组合图像、文本、视频输入,保持场景控制一致性。3. 现实世界知识:调用Gemini在历史、生物等领域的知识来构建视频,无需复杂的描述。4. 文字与动作同步:通过简单提示词,将文字和图形直接连接到视频动作中。

Q文章提到谷歌展示了哪三个Demo APP来体现两个新模型的协同工作流程?

A谷歌展示了三个Demo APP来体现图像生成模型Nano Banana 2 Lite与视频生成模型Gemini Omni Flash的协同工作流程:1. Anywhere:用户上传自拍,NB2 Lite将用户P到地标景点照片中,Omni Flash再将静态景点图片转为动态短片。2. Space Lift:上传房间照片,NB2 Lite生成装修方案图,Omni Flash将选定方案生成电影级空间漫游视频。3. Omni product studio:上传产品白底图,NB2 Lite生成场景化商品图,Omni Flash将其转为电商短视频。

Q文章认为谷歌在多模态领域的优势是什么?

A文章认为,在当前众多厂商“死磕”代码能力(Coding)的情况下,谷歌在多模态领域持续投入并推出了Gemini Omni Flash和Nano Banana 2 Lite等模型,展示了多模态在电商、装修、短视频等垂直场景落地的潜力。凭借其自身产品生态(如Stitch、Pixel、NotebookLM)和安卓生态的加持,商业化前景明确。作者认为,在多模态这张牌桌上,谷歌可能是“唯一能组齐一套牌的玩家”。

Leituras Relacionadas

QCP Capital Analysts Explained Why Bitcoin Stalled at $63,000

QCP Capital's latest report explains that Bitcoin has stalled around $63,000, testing the lower boundary of its recent trading range ($62,500–$64,000). The retreat wasn't due to a single catalyst but stems from several cross-market pressures: heightened geopolitical risks, rising oil prices, uncertainty around global liquidity, and a muted reaction to positive U.S. macroeconomic data. While U.S. inflation and jobs data softened, reducing expectations of imminent Fed tightening, Bitcoin's price response has been subdued. This highlights the importance of other factors like trader positioning, liquidity conditions, and overall risk appetite. The regulatory landscape is progressing on two tracks: congressional legislation (like the CLARITY Act) is delayed, but SEC rulemaking continues. Corporate treasury activity, exemplified by Strategy's new Bitcoin monetization program, now creates a potential two-way flow, adding complexity to supply dynamics. Geopolitical tensions around the Strait of Hormuz keep oil prices elevated, impacting crypto indirectly through inflation expectations and financial conditions. Historical seasonality shows August and September are typically weak for Bitcoin, but this is descriptive, not predictive, in today's ETF-driven market. In conclusion, Bitcoin has shown resilience by absorbing multiple negative forces without a decisive breakdown. However, it has failed to convert improving macro fundamentals into sustained upward momentum. The market awaits key events like the U.S. PCE data, the Jackson Hole symposium, and the September FOMC meeting for a potential catalyst to break the current range.

cryptonews.ruHá 1m

QCP Capital Analysts Explained Why Bitcoin Stalled at $63,000

cryptonews.ruHá 1m

Did DeepSeek Intentionally Make an Unfriendly Agent?

DeepSeek is shifting its strategy from being a low-cost model provider to building a foundational ecosystem for AI agents. This pivot is marked by two recent moves: increasing the price of its flagship V4 Pro model and launching the DeepSeek Harness agent framework. Previously, DeepSeek's success was built on creating powerful, affordable models and open-sourcing them to drive adoption. Now, the focus is moving "beyond the model" to the runtime environment where agents operate. Harness is key to this new approach. Unlike integrated agent products from competitors, Harness deconstructs the agent stack—separating the model, tools, skills, workflows, and UI into modular, replaceable components. Its core innovation is the Cordis kernel, which enables dynamic loading, unloading, and management of plugins, allowing agents to modify their own capabilities during execution. Significantly, Harness does not lock users into DeepSeek's models; it allows integration with third-party models. This suggests DeepSeek's ambition is not to control the model layer but to establish the standard runtime environment for future agents—akin to an operating system for the agent era. The V4 Pro price increase aligns with this shift: the model transitions from a loss-leading acquisition tool to a core revenue engine within a broader ecosystem. Harness is currently complex and developer-focused, prioritizing ecosystem builders over end-users. Its success will be measured not by immediate user numbers, but by the plugins, agents, and platforms built upon it. DeepSeek's long-term bet is that controlling the agent runtime will be more valuable than merely providing the underlying models, potentially redefining how AI agents are created and evolved.

marsbitHá 37m

Did DeepSeek Intentionally Make an Unfriendly Agent?

marsbitHá 37m

Metrics Ventures Market Observation: Talk is Cheap

This monthly market analysis extends its timeline to incorporate critical July comments from the Fed Chair, noting that bond markets have already priced in perceived policy shortcomings. The report observes a growing divergence: equity markets, after some deleveraging, continue a "trust-based" rally, while bond and currency markets signal persistent distrust. Precious metals bottoming suggests a central bank consensus that the era of "competitive currency devaluation" is ending, with verbal interventions losing power. Looking forward to Q3-Q4, the analysis remains bullish on supply-constrained global resources like copper and power, as well as gold, which prices ongoing monetary失信. It argues that digital assets are unlikely to see major outperformance until excess liquidity is released and AI growth rates are fully priced. Key market views include: 1. Commodities like gold remain primary liquidity absorbers over Bitcoin, with recent consolidation seen as healthy. 2. The bull trend for RMB-denominated assets (e.g., STAR 50 Index) is firmly established. 3. Key resource country indices and currencies are near inflection points, with spot copper already at new highs. The report suggests resource equities, particularly in China's market, are at the end of their consolidation phase, offering attractive valuations with embedded optionality on rising metal prices. It highlights the predictive significance of recent US-Japan FX interventions and Treasury-Fed dynamics, suggesting a shift towards less communication and data management to maintain stability. A long position in resource assets is presented as a positive expected-value strategy over a multi-year horizon.

marsbitHá 2h

Metrics Ventures Market Observation: Talk is Cheap

marsbitHá 2h

The Value, Growth, and Risks of Prediction Markets

The article discusses the explosive growth and complex nature of prediction markets in the United States, focusing on their value, risks, and regulatory challenges. Key drivers include sports betting, which fueled a 1795% year-over-year trading volume increase in Q2 2026, with platforms like Kalshi and Polymarket dominating. These markets offer significant potential as superior hedging tools for businesses and more efficient price discovery mechanisms than traditional polls or derivatives for events like Fed rate decisions, elections, or GPU prices. They also serve as customer acquisition channels for platforms like Robinhood. However, major challenges persist. A central conflict exists between federal regulators (CFTC), which classify event contracts as derivatives under its jurisdiction, and state authorities, which view sports-related contracts as illegal gambling, leading to numerous lawsuits. There is significant consumer protection risk: data shows most retail users lose money to professional traders, platforms market high-risk products like parlays, and protections (e.g., age limits, addiction help) are weaker than in state-regulated sports betting. The "self-certification" process allows rapid contract launches but creates regulatory uncertainty. The author argues prediction markets are fundamentally "better markets" but currently fail to provide adequate consumer safeguards. Proposals include integrating protection mechanisms (cooling-off periods, position limits), aligning risk warnings with product risks, creating paths from speculation to long-term investing, and applying specific consumer protections and taxes to sports contracts. The piece concludes by urging stakeholders to address these issues before a potential regulatory backlash undermines the markets' legitimate utility.

marsbitHá 2h

The Value, Growth, and Risks of Prediction Markets

marsbitHá 2h

Trading

Spot

Artigos em Destaque

Como comprar 4

Bem-vindo à HTX.com!Tornámos a compra de 4 (4) simples e conveniente.Segue o nosso guia passo a passo para iniciar a tua jornada no mundo das criptos.Passo 1: cria a tua conta HTXUtiliza o teu e-mail ou número de telefone para te inscreveres numa conta gratuita na HTX.Desfruta de um processo de inscrição sem complicações e desbloqueia todas as funcionalidades.Obter a minha contaPasso 2: vai para Comprar Cripto e escolhe o teu método de pagamentoCartão de crédito/débito: usa o teu visa ou mastercard para comprar 4 (4) instantaneamente.Saldo: usa os fundos da tua conta HTX para transacionar sem problemas.Terceiros: adicionamos métodos de pagamento populares, como Google Pay e Apple Pay, para aumentar a conveniência.P2P: transaciona diretamente com outros utilizadores na HTX.Mercado de balcão (OTC): oferecemos serviços personalizados e taxas de câmbio competitivas para os traders.Passo 3: armazena teu 4 (4)Depois de comprar o teu 4 (4), armazena-o na tua conta HTX.Alternativamente, podes enviá-lo para outro lugar através de transferência blockchain ou usá-lo para transacionar outras criptomoedas.Passo 4: transaciona 4 (4)Transaciona facilmente 4 (4) no mercado à vista da HTX.Acede simplesmente à tua conta, seleciona o teu par de trading, executa as tuas transações e monitoriza em tempo real.Oferecemos uma experiência de fácil utilização tanto para principiantes como para traders experientes.

793 Visualizações TotaisPublicado em {updateTime}Atualizado em 2026.06.02

Como comprar 4

Discussões

Bem-vindo à Comunidade HTX. Aqui, pode manter-se informado sobre os mais recentes desenvolvimentos da plataforma e obter acesso a análises profissionais de mercado. As opiniões dos utilizadores sobre o preço de 4 (4) são apresentadas abaixo.

活动图片