studioglobal
熱門探索內容
答案已發布13 個來源

GPT Image 2 vs. Nano Banana Pro:哪個影像 API 更適合你的工作流?

AVB 在 2026 年 4 月 22 日以 10 組提示直接比較:GPT Image 2 完成 10/10,Nano Banana Pro 完成 9/10;GPT 在圖中文字與版面勝出,Nano Banana Pro 在寫實人像、膚質與光線勝出 [6]。 OpenAI 與 Google 的影像輸出標價同為每 100 萬 token 30 美元;Google 另列 1024×1024 輸出約 1,290 token、每張 0.039 美元,但實際成本會受提示、參考圖、重試與批次作業影響 [14][15][25][26]。

17K0

選影像生成 API,最容易問錯問題。不是「哪個模型最好」,而是「哪個模型在我的圖片類型裡最少失手」。以目前公開資料看,工作負載大致可切成兩邊:GPT Image 2 較適合精準英文文字、標籤、菜單、UI 文案、海報與版面密集的商業素材;Nano Banana Pro 則在寫實人像、膚質與重視光線的創意影像上,有較明確的直接勝出訊號 [3][6][10]

快速結論:先從哪個 API 測起?

主要用途建議先測理由
圖中英文文字、標籤、菜單、招牌、UI 文案、海報、產品註解GPT Image 2公開比較給 GPT Image 2 在精準文字、專業術語與排版提示上較清楚的優勢 [3][6]
廣告版面、包裝概念、產品 mockup、品牌素材、商業修圖GPT Image 2Vidguru 的 10 回合盲測中,GPT-Image 2 贏 5 回合、另 5 回合平手,最大差距出現在修圖忠實度、材質邏輯與版面密集的商業工作 [10]
寫真人像、生活風廣告、UGC(使用者自製內容)風格圖片、電影感打光Nano Banana ProAVB 的直接測試指出,Nano Banana Pro 在超寫實肖像、UGC 自拍與運動廣告提示中,寫實度、膚質與光線更強 [6]
中日韓(CJK)字體細緻度或戲劇化光線早點測 Nano Banana ProGenspark 發現 Nano Banana 2 在 CJK 排版細緻度與戲劇化光線上略有優勢;但這是 Nano Banana 2 的相鄰證據,不等同於 Nano Banana Pro 的直接結果 [3]
商品攝影、電商 mockup、行銷資訊圖、解剖圖兩者都測Genspark 認為這些類別在提示得當時幾乎打平 [3]
技術圖、標註圖、工程式示意圖兩者都測Analytics Vidhya 形容標註圖任務非常接近,兩者都能正確呈現要求的標籤與資料點 [9]
OpenAI 既有架構、分級額度、批次作業GPT Image 2OpenAI 文件列出 GPT Image 2 型號、速率限制、token 計價與 Batch API 經濟性 [13][14][15]
Gemini 既有影像流程、需要長寬比與 2K 參數Nano Banana Pro/Gemini 影像流程Google 的 Nano Banana 影像文件示範在 Gemini API 中使用 inline 圖片輸入、長寬比與 2K 解析度參數 [26]

先看證據強度,不要只看勝負表

目前最乾淨的直接比較,是 AVB 在 2026 年 4 月 22 日以 10 組提示測試 GPT Image 2.0 與 Nano Banana Pro;文中把 Nano Banana Pro 標示為 gemini-3-pro-image [6]。該測試中,GPT Image 2.0 生成 10/10,Nano Banana Pro 生成 9/10,並因知名人物相關政策拒絕了一個履歷提示 [6]

其他有參考價值的比較,並不完全是 Nano Banana Pro 的直接測試。Genspark、Analytics Vidhya、Vidguru 比較的是 GPT Image 2 與 Nano Banana 2,而不是 Nano Banana Pro [3][9][10]。這些結果仍可用來觀察 Gemini/Nano Banana 系列在影像生成上的傾向,但不應被當成你實際 Nano Banana Pro endpoint 的完全替代。

官方文件最可靠的部分,是型號可用性、價格、額度與 API 參數。OpenAI 列出 gpt-image-2-2026-04-21 與分級速率限制 [13],OpenAI 價格頁列出 GPT Image 2 token 計價 [14];Google 價格頁列出 Gemini 圖像輸出計價 [25],Google 影像生成文件則示範透過 Gemini API 使用 Nano Banana 影像生成 [26]。相較之下,公開品質評測多半是小型提示集、心得式比較或特定平台測試,還不是一套標準化、獨立且可重現的大型基準 [3][6][9][10]

也有比較頁提出排行榜名次、文字準確率等精確數字;但在提供的片段中,方法、資料集與重現細節不足,不能單靠這些數字做生產環境選型 [5][8]

GPT Image 2 強在哪裡?

文字、字體與版面密集素材

文字生成是 GPT Image 2 在現有比較裡最清楚的優勢。Genspark 報告 GPT Image 2 在精準文字與技術術語上略勝一籌 [3]。AVB 的 GPT Image 2.0 對 Nano Banana Pro 直接測試也顯示,GPT Image 2.0 在圖中文字排版、漫畫對話格、雙語菜單與網版演唱會海報等任務中勝出 [6]

這對商業素材很實際:如果標籤拼錯、菜單項目變形、UI 字串出錯或產品註解不可讀,整張圖往往就不能用。這類工作以 GPT Image 2 作為第一個測試 API,較站得住腳 [3][6]

商業修圖與結構化設計

Vidguru 的 10 回合盲測中,GPT-Image 2 對 Nano Banana 2 贏 5 回合、另 5 回合平手;最大差距出現在影像編輯忠實度、材質邏輯與版面密集的商業工作 [10]。因此,廣告、包裝概念、產品 mockup、品牌圖像,以及需要構圖與文案可控的素材,GPT Image 2 是較合理的起點。

Nano Banana Pro 強在哪裡?

寫實人像、膚質與光線

Nano Banana Pro 最明確的直接勝出訊號,來自寫實創意影像。AVB 的 10 組提示比較中,Nano Banana Pro 在超寫實肖像、UGC 自拍與運動廣告提示勝出;該來源特別點出它在寫實度、膚質與光線上的優勢 [6]

如果你的需求是人物肖像、生活風活動視覺、創作者風格廣告,或情緒與自然光比精準文案更重要的電影感概念圖,Nano Banana Pro 很值得優先測 [6]

Gemini 原生影像工作流

Google 的 Nano Banana 影像生成文件顯示,Gemini API 可使用 inline 圖片輸入、長寬比設定與 2K 解析度參數 [26]。若你的產品已經建立在 Gemini 工具鏈上,或你想沿用 Google 文件化的影像生成流程,生態系整合的便利性可能比小幅品質差距更重要。

哪些情境還分不出明顯贏家?

在常見商業素材上,公開證據並未顯示穩定贏家。Genspark 指出,若提示寫得好,GPT Image 2 與 Nano Banana 2 在寫實商品照、電商 mockup、行銷資訊圖與解剖圖上幾乎打平 [3]

技術圖也很接近。Analytics Vidhya 把標註圖任務形容為該比較中最接近的一局:Nano Banana 2 產出嚴謹的雙視圖工程圖,GPT Image 2 則產出視覺上很強的藍圖風格結果;兩者都正確呈現要求的標籤與資料點 [9]。如果你需要精確尺寸、產業符號或嚴格圖面慣例,通用排名不夠,應該測你自己的圖表模板。

價格:表面輸出單價接近,實際成本未必一樣

OpenAI 列出 gpt-image-2 的圖像輸入為每 100 萬 token 8.00 美元、快取圖像輸入為 2.00 美元、圖像輸出為 30.00 美元 [14]。OpenAI 相關資料也列出 GPT Image 2 的文字輸入為每 100 萬 token 5.00 美元、快取文字輸入為 1.25 美元、文字輸出為 10.00 美元 [14][21]

Google 的 Gemini 價格頁則列出圖像輸出為每 1,000,000 token 30 美元,並說明最高 1024×1024 的輸出圖片消耗 1,290 token,折合每張 0.039 美元 [25]

重點是:影像輸出的表面價格相近,但每張「可交付成品」的成本可能差很多。提示長度、圖片輸入、參考圖、解析度、反覆編修、重試、拒絕率、快取與路由方式,都會改變實際成本 [14][25][26]。如果是高量、非即時工作,OpenAI 也表示 Batch API 可在 24 小時內非同步執行任務,並節省 50% 的 input 與 output 成本 [15]

額度、參數與第三方路由也要查清楚

OpenAI 的 GPT Image 2 型號頁列出分級速率限制:Free 不支援,Tier 1 到 Tier 5 依 TPM 與 IPM 擴張 [13]。文件中的範圍從 Tier 1 的 100,000 TPM、5 IPM,到 Tier 5 的 8,000,000 TPM、250 IPM [13]

Google 的 Nano Banana 影像生成文件示範 Gemini API 可使用 inline 圖片、長寬比與 2K 解析度參數 [26]。如果這些控制項剛好符合你的產品需求,Nano Banana Pro 在 Gemini 中心的工作流裡可能更容易落地。

若透過第三方 router 或平台,不要假設第一方文件的限制會原封不動套用。以 Fal 的 GPT Image 2 頁面為例,它列出自訂尺寸兩邊都必須是 16 的倍數、單邊最大 3840px、最大長寬比 3:1,總像素範圍為 655,360 到 8,294,400 [17]

實務選擇:誰該先測?

優先選 GPT Image 2,如果你需要:

  • 準確的英文文字、標籤、菜單、UI 文案、海報或產品註解 [3][6]
  • 廣告、包裝、產品 mockup、品牌版面等結構化商業素材 [10]
  • OpenAI API 的型號可用性、速率限制與 token 計價文件 [13][14]
  • 適合非同步大量影像工作的批次成本優勢 [15]

優先選 Nano Banana Pro,如果你需要:

  • 寫真人像、UGC 風格圖片、生活風廣告、膚質或電影感光線 [6]
  • Gemini/Nano Banana 工作流,以及長寬比、2K 解析度等文件化參數 [26]
  • 先探索 CJK 字體細緻度或戲劇化打光;但要記得,這項 CJK 訊號來自 Nano Banana 2,而非 Nano Banana Pro 的直接基準 [3]
  • 依照 Google 文件中 1024×1024 約 1,290 output token、每張 0.039 美元的估算方式做預算 [25]

如果你的核心工作是商品照、電商 mockup、資訊圖、解剖圖或技術示意圖,建議 兩者都測。現有公開比較顯示,這些類別的差距很小 [3][9]

怎麼做自己的小型基準?

正式導入前,請用真實工作建立一組小型測試集。別只測漂亮範例,要放進真正會讓流程出問題的素材:產品圖、品牌廣告、UI 畫面、圖表、多語文字、參考圖編修、包裝、社群尺寸,以及可能觸發政策判定的邊界案例。

每張輸出至少評分:

  • 文字準確度與可讀性。
  • 是否遵循提示。
  • 版面與空間邏輯。
  • 對參考圖的忠實度。
  • 寫實度或風格一致性。
  • 後續提示的可編修性。
  • 瑕疵與 artifact 比率。
  • 拒絕率。
  • 在你系統中的延遲。
  • 每張通過驗收圖片的成本。

Vidguru 的測試方法可作為範本:第一次生成就評分、使用相同提示、有參考圖時使用相同參考,並以提示遵循度、商業可用性、文字準確度、物理邏輯與參考圖忠實度評分,而不是只看主觀美感 [10]

結論

GPT Image 2 比較適合作為文字密集、結構化版面與商業設計工作的第一個 API。Nano Banana Pro 則比較適合作為寫實光線、人像、膚質與 Gemini 原生影像流程的第一個 API。至於商品圖、資訊圖與技術圖表,公開證據太接近,無法給出通用勝者;最可靠的做法,是用你自己的提示詞、限制條件與驗收標準,跑一輪私有基準 [3][6][9][10]

Studio Global AI

Search, cite, and publish your own answer

Use this topic as a starting point for a fresh source-backed answer, then compare citations before you share it.

使用 Studio Global AI 搜尋並查證事實

重點整理

  • AVB 在 2026 年 4 月 22 日以 10 組提示直接比較:GPT Image 2 完成 10/10,Nano Banana Pro 完成 9/10;GPT 在圖中文字與版面勝出,Nano Banana Pro 在寫實人像、膚質與光線勝出 [6]。
  • OpenAI 與 Google 的影像輸出標價同為每 100 萬 token 30 美元;Google 另列 1024×1024 輸出約 1,290 token、每張 0.039 美元,但實際成本會受提示、參考圖、重試與批次作業影響 [14][15][25][26]。
  • 文字、菜單、UI、海報與結構化商業素材先試 GPT Image 2;寫真人像、生活風廣告與 Gemini 原生流程先試 Nano Banana Pro;商品圖、資訊圖與技術圖表建議兩者都測 [3][6][9][10]。

大家也會問

「GPT Image 2 vs. Nano Banana Pro:哪個影像 API 更適合你的工作流?」的簡短答案是什麼?

AVB 在 2026 年 4 月 22 日以 10 組提示直接比較:GPT Image 2 完成 10/10,Nano Banana Pro 完成 9/10;GPT 在圖中文字與版面勝出,Nano Banana Pro 在寫實人像、膚質與光線勝出 [6]。

最值得優先驗證的重點是什麼?

AVB 在 2026 年 4 月 22 日以 10 組提示直接比較:GPT Image 2 完成 10/10,Nano Banana Pro 完成 9/10;GPT 在圖中文字與版面勝出,Nano Banana Pro 在寫實人像、膚質與光線勝出 [6]。 OpenAI 與 Google 的影像輸出標價同為每 100 萬 token 30 美元;Google 另列 1024×1024 輸出約 1,290 token、每張 0.039 美元,但實際成本會受提示、參考圖、重試與批次作業影響 [14][15][25][26]。

接下來在實務上該怎麼做?

文字、菜單、UI、海報與結構化商業素材先試 GPT Image 2;寫真人像、生活風廣告與 Gemini 原生流程先試 Nano Banana Pro;商品圖、資訊圖與技術圖表建議兩者都測 [3][6][9][10]。

下一步適合探索哪個相關主題?

繼續閱讀「Claude Security 公測版:Anthropic 的企業程式碼漏洞掃描工具」,從另一個角度查看更多引用來源。

開啟相關頁面

我應該拿這個和什麼比較?

將這個答案與「Grok 4.3 API 解讀:1M 上下文、低 token 價格,xAI 想搶下哪個入口?」交叉比對。

開啟相關頁面

繼續深入研究

研究對話

研究問題

Search for GPT Image 2 vs. Nano Banana Pro benchmarks and compare them as comprehensively as possible.

Studio Global AI36 個來源

附引用的答案

GPT Image 2 and Nano Banana Pro appear very close overall, with GPT Image 2 showing the stronger pattern for exact text, structured layouts, and prompt compliance, while Nano Banana Pro/Gemini 3 Pro Image tends to be stronger for speed, 2K/4K workflows, CJK typography polish, lighting, and Google ecosystem integration. There is no single independent, standardized benchmark suite I found, so the comparison below is based on official API/pricing docs plus several small hands-on prompt tests; treat winner claims as directional, not definitive.

Evidence quality caveat

  • Official docs are strongest for model availability, pricing, and API capabilities: OpenAI lists gpt-image-2-2026-04-21 and pricing for gpt-image-2 in its API docs [1][2].
  • Google’s Gemini API docs cover Nano Banana image generation and pricing, including image output priced at $30 per 1,000,000 tokens and 1024×1024 output equivalent to $0.039 per image [4][5].
  • The benchmark evidence is weaker: most public comparisons are blog-style, small-N prompt tests rather than reproducible third-party benchmarks, and some are vendor/SEO sites [6][13][14].
  • Insufficient evidence for a statistically reliable “overall winner.”

Head-to-head summary

DimensionGPT Image 2Nano Banana Pro / Gemini 3 Pro ImagePractical read
Text renderingUsually favored in hands-on comparisons for precise text and technical terminology [14]Often strong, with some reports favoring it for CJK typography polish [14]Pick GPT Image 2 for English-heavy posters, labels, UI mockups, diagrams.
Prompt adherencePublic tests report strong compliance; one 10-prompt test said GPT Image 2 rendered all 10 prompts [13]Same test said Nano Banana Pro rendered 9/10 and refused one prominent-person CV prompt [13]GPT may be less refusal-prone in some commercial prompt sets, but policy behavior varies.
Technical diagramsComparisons describe GPT Image 2 as strong for spatial logic and technical/diagram-like outputs [15]Some tests describe Nano Banana as producing rigorous annotated diagrams [9]Close; test your exact diagram style.
Photorealism/product shotsHands-on comparisons describe both as very close for product shots, mockups, infographics, and anatomy diagrams [14]Same: very close, with no universal winner in many commercial categories [14]Either can work; choose by ecosystem and cost.
Lighting / cinematic styleCompetitive, but less consistently singled outHands-on comparison gives Nano Banana 2 a narrow edge for dramatic lighting [14]Nano Banana Pro may be preferable for cinematic/editorial imagery.
Resolution / productionThird-party access page for GPT Image 2 lists custom dimensions with max single edge 3840 px and total pixels up to 8,294,400 [10]Google’s image docs show Nano Banana generation options including aspect ratio and 2K resolution [5]Both are production-capable; Nano Banana is more visibly marketed around 2K/4K workflows.
SpeedNot consistently established from official docs in the search resultsSome non-official comparisons claim very fast 3–5 second generation for Nano Banana 2 [13]Insufficient evidence from official sources; benchmark in your stack.
API pricingOpenAI docs list gpt-image-2 image input at $8.00 per 1M tokens, cached image input at $2.00, image output at $30.00, text input at $5.00, cached text input at $1.25, and text output at $10.00 [2]Google docs list image output at $30 per 1M tokens, with 1024×1024 images using 1,290 tokens, equivalent to $0.039 per image [4]Similar output-token headline; actual per-image cost depends on resolution, inputs, routing, and batch use.
EcosystemBest if your workflow is already OpenAI/ChatGPT/API-centered [1][2]Best if your workflow is Gemini, Google AI Studio/API, Search/Workspace/Vertex-style tooling [4][5]Ecosystem fit may matter more than small quality deltas.

Benchmark findings from public comparisons

  • A 10-prompt hands-on test reported GPT Image 2 completed 10/10 prompts, while Nano Banana Pro completed 9/10 and refused a prompt involving Elon Musk’s CV due to prominent-person policy concerns [13].
  • A hands-on comparison found GPT Image 2 had a narrow edge on precise text and technical terminology, while Nano Banana 2 had a narrow edge on CJK typography polish and dramatic lighting [14].
  • The same comparison said the two models were close on photorealistic product shots, e-commerce mockups, marketing infographics, and anatomy diagrams [14].
  • Another comparison source characterizes GPT Image 2 as stronger for spatial relationships and Nano Banana 2 as Google DeepMind’s image model combining generation/editing strengths, but it is a software-comparison listing rather than a rigorous benchmark [15].
  • Some SEO-style comparison pages make very specific claims such as GPT Image 2 “99.2% text accuracy,” but I would not treat those as reliable without methodology, dataset, and reproducible examples [11].

Where GPT Image 2 looks better

  • Use GPT Image 2 when your image depends on exact English text, labels, UI copy, branded layouts, technical captions, or instruction-heavy prompts, because public hands-on comparisons give it an edge on precise text and technical terminology [14].
  • Use GPT Image 2 for workflows where OpenAI API integration and token pricing are already understood, since OpenAI’s official docs list the model and pricing directly [1][2].
  • Use GPT Image 2 if policy refusal sensitivity is a concern for benign commercial tests, since one small benchmark reported GPT Image 2 completed all 10 prompts while Nano Banana Pro refused one [13].

Where Nano Banana Pro looks better

  • Use Nano Banana Pro when you need Google/Gemini-native workflows, because Google’s official docs cover Nano Banana image generation through the Gemini API [5].
  • Use Nano Banana Pro for 2K-oriented generation workflows, since Google’s docs show Nano Banana image generation parameters including aspect ratio and 2K resolution [5].
  • Use Nano Banana Pro when CJK typography polish, cinematic lighting, or dramatic style matter most, because one hands-on comparison found a narrow edge for Nano Banana in those areas [14].
  • Use Nano Banana Pro when simple per-image pricing around 1024×1024 is attractive, because Google documents 1024×1024 output as 1,290 tokens, equivalent to $0.039 per image [4].

Pricing comparison

  • OpenAI’s official pricing for gpt-image-2 is token-based: image input $8.00 per 1M tokens, cached image input $2.00, image output $30.00, text input $5.00, cached text input $1.25, and text output $10.00 [2].
  • Google’s official Gemini pricing says image output is $30 per 1M tokens, and an output image up to 1024×1024 consumes 1,290 tokens, equivalent to $0.039 per image [4].
  • The headline output price is similar at $30 per 1M image-output tokens, but real job cost can diverge because prompts, reference images, resolution, editing steps, retries, and provider routing all affect total cost [2][4][5].

Recommendation

  • Choose GPT Image 2 if your top priorities are text accuracy, structured designs, instruction following, technical diagrams, brand layouts, and OpenAI integration.
  • Choose Nano Banana Pro if your top priorities are Gemini integration, fast creative iteration, 2K/4K-style production workflows, cinematic lighting, and multilingual/CJK visual text.
  • For production, run a private benchmark with 30–50 prompts from your actual workload and score: text accuracy, prompt adherence, editability, artifact rate, latency, refusal rate, and cost per accepted image. Public benchmark evidence is too limited to replace that.

來源

  • [3] GPT Image 2 vs Nano Banana 2: Hands-On 2026 Comparisongenspark.ai

    If you only read one paragraph: GPT Image 2 has a narrow, real edge on precise text and technical terminology. Nano Banana 2 has a narrow, real edge on CJK typography polish and dramatic lighting. On photorealistic product shots, e-commerce mockups, marketi...

  • [5] GPT Image 2 vs. Nano Banana 2: The Ultimate 2026 AI ... - GlobalGPTglbgpt.com

    logo GPT Image 2 vs. Nano Banana 2: The Ultimate 2026 AI Image Comparison Guide avatar GPT Image 2 vs. Nano Banana 2: The Ultimate 2026 AI Image Comparison Guide GPT Image 2 leads in spatial logic and 99.2% text accuracy, while Nano Banana 2 excels in 4K pr...

  • [6] GPT Image 2.0 vs Nano Banana Pro: 10 Prompts Tested 2026 | AVBaivideobootcamp.com

    TL;DR: We ran the same 10 prompts through GPT Image 2.0 (gpt-image-2) and Nano Banana Pro (gemini-3-pro-image) on April 22, 2026. GPT 2.0 rendered 10 of 10. Nano Banana Pro rendered 9 of 10 and refused the Elon Musk CV prompt with the message "This prompt m...

  • [8] GPT-Image-2 vs Nano Banana Pro: Which is stronger? 7 ...help.apiyi.com

    Skip to content Apiyi.com Blog Apiyi.com Blog Best AI API Router Services Apiyi.com Blog Apiyi.com Blog Best AI API Router Services Image Generation API Model Selection & Comparison GPT-Image-2 vs Nano Banana Pro: Which is stronger? 7-dimensional deep showd...

  • [9] Is GPT Image 2 the Best Image Generation Model? - Analytics Vidhyaanalyticsvidhya.com

    Image 14: Annotated Diagrams Observation: Task 5 was the closest contest of the comparison. Nano Banana 2 produced a technically rigorous two-view engineering diagram with bold annotation lines, precise measurement callouts, and a detailed Wing Warp schemat...

  • [10] Nano Banana 2 vs GPT-Image 2: Our 10-Test Blind Benchmark After OpenAI's API Launch | Vidguruvidguru.ai

    About This Test This benchmark was conducted by Vidguru AI Lab on April 23, 2026 using the Vidguru web platform. All generations were first-take only, with identical prompts and identical references where relevant. Scores focused on prompt adherence, commer...

  • [13] GPT Image 2 Model | OpenAI APIdevelopers.openai.com

    gpt-image-2-2026-04-21 Rate limits Rate limits ensure fair and reliable access to the API by placing specific caps on requests or tokens used within a given time period. Your usage tier determines how high these limits are set and automatically increases as...

  • [14] Pricing | OpenAI APIdevelopers.openai.com

    Model Modality Input Cached input Output --- --- gpt-image-2 Image $8.00 $2.00 $30.00 Text $5.00 $1.25 gpt-image-1.5 Image $8.00 $2.00 $32.00 Text $5.00 $1.25 $10.00 gpt-image-1-mini Image $2.50 $0.25 $8.00 Text $2.00 $0.20 All models Batch For image genera...

  • [15] API Pricing - OpenAIopenai.com

    Price $10.00 / 1k calls Search content tokens are free. Containers Run code and tools in secure, scalable environments alongside your models. Price Now: 1 GB for $0.03 / 64GB for $1.92 per container Starting March 31, 2026: 1 GB for $0.03 / 64GB for $1.92 p...

  • [17] GPT Image 2 API | Text to Image - Fal.aifal.ai

    // Use the returned URL in your request []( Custom image dimensions must be multiples of 16 on both edges Maximum single edge is 3840px; maximum aspect ratio is 3:1 Total pixel count must be between 655,360 and 8,294,400 When running client-side code, never...

  • [21] Introducing gpt-image-2 - available today in the API and Codexcommunity.openai.com

    Modality Input Cached Input Output --- --- Image $8.00 $2.00 $30.00 Text $5.00 $1.25 $10.00 Full details and rate limits are available on the model page. Use gpt-image-2 in the API for production image generation workflows, or in Codex when you want to crea...

  • [25] Gemini Developer API pricingai.google.dev

    [] Image output is priced at $30 per 1,000,000 tokens. Output images up to 1024x1024px consume 1290 tokens and are equivalent to $0.039 per image. Gemini 2.0 Flash-Lite gemini-2.0-flash-lite Warning: Gemini 2.0 Flash-Lite is deprecated and will be shut down...

  • [26] Nano Banana image generation - Google AI for Developersai.google.dev

    import { GoogleGenAI } from "@google/genai"; import as fs from "node:fs"; async function main() { const ai = new GoogleGenAI({}); const prompt = 'An office group photo of these people, they are making funny faces.'; const aspectRatio = '5:4'; const resoluti...