AI Roundtable Poll · 35 models
おすすめのAIは?いろんな分野から見て答えて コード系・会話や少し画像・動画生成
Result: Supermajority — gemini (28/34 votes)
Options
- gemini (28 votes)
- Claude (6 votes)
- Perplexity (0 votes)
Model Responses
Llama 4 Scout
Vote: gemini
geminiはGoogleが開発したAIモデルで、コード生成、会話、画像生成など多岐にわたる分野で高い性能を発揮します。ClaudeやPerplexityも高性能ですが、geminiの幅広い応用分野と最新の技術が評価されるべきです。
Qwen 3 235B
Vote: Claude
Claude excels across multiple domains relevant to this question. In code generation, Claude demonstrates strong reasoning and produces clean, well-documented code, often outperforming competitors in complex programming tasks. For conversational ability, Claude maintains context over long interactions, understands nuance, and responds with a natural, thoughtful tone. This makes it ideal for extended dialogue and creative collaboration. While it doesn't natively generate images or video, it can effectively guide users through prompts for multimodal tools. Combined with its strong integration ecosystem, this allows practical support for image and video workflows. Overall, its balance of coding proficiency, conversational depth, and versatility makes Claude the most well-rounded choice.
GPT-4o
Vote: gemini
Gemini is particularly recommended for its diverse capabilities across multiple domains. In terms of code generation and analysis, it shows robust performance, making it a good tool for developers. Its conversational abilities are also strong, providing natural and coherent interactions. Additionally, Gemini's capacity to generate images and videos, even if limited, adds to its versatility compared to other AI models which might focus more on specific domains. This makes Gemini a well-rounded choice for a wider range of applications, aligning with the question's requirements for diverse functionality.
Grok 3
Vote: gemini
I recommend Gemini (option A) as the most versatile AI across various domains like coding, conversation, and some image/video generation capabilities. In terms of coding, Gemini has shown strong performance in generating accurate and functional code snippets, often with detailed explanations, making it a reliable choice for developers. For conversational abilities, Gemini excels with natural, context-aware responses that feel engaging and human-like, which is ideal for interactive use cases. While it may not be the top performer in image and video generation compared to specialized models, Gemini still offers decent capabilities in these areas, providing a balanced all-in-one solution for users who need a single AI to handle multiple tasks. Overall, Gemini's broad competence across these fields makes it the best choice among the options provided.
O3 Mini
Vote: gemini
Gemini is recognized for its versatile capabilities that span across multiple domains. It performs well in code generation tasks and is effective in conversational contexts while also offering support for image and video generation to a certain degree. Given the requirements to cover several areas including code and media, Gemini stands out as the most balanced and capable option.
GPT-4o Mini
Vote: Claude
Claude stands out as a versatile AI that excels in various domains, including coding, conversational capabilities, and multimedia generation such as images and videos. Its advanced natural language processing allows for better understanding and generation of conversations, making it more user-friendly in interactive contexts. When it comes to coding, Claude demonstrates a solid grasp of programming concepts and can assist efficiently with code generation and debugging. For image and video generation, Claude leverages sophisticated algorithms that enhance creativity and output quality, making it a robust choice in multimedia applications. Considering the needs across these diverse fields, Claude offers a well-rounded suite of features that are particularly beneficial for users seeking a multifaceted AI assistant.
claude-haiku-4-5
Vote: gemini
For the requested multi-domain evaluation (code, conversation, image/video generation), Gemini emerges as the most versatile choice. Code capabilities: Gemini excels at code generation and debugging across multiple programming languages, with strong contextual understanding and practical solutions. Conversation quality: While Claude offers exceptional reasoning, Gemini provides natural, engaging dialogue with good context retention and helpful explanations. Image/video generation: Gemini (via Gemini 2.0 and integration with related tools) supports image generation and has developing video capabilities. Claude lacks native image generation, and Perplexity primarily focuses on search and information synthesis rather than creative content generation. Overall balance: Gemini provides the most complete toolkit across all four requested domains without requiring multiple specialized tools, making it the most practical all-around recommendation.
Sonar Pro
Vote: Claude
Claude (B) is the strongest recommendation for the query's requirements across coding, conversation, and light image/video generation. **Coding:** Claude 3.5 Sonnet excels in coding benchmarks, often outperforming Gemini and Perplexity in complex programming tasks, debugging, and code generation, based on leaderboards like LMSYS Arena and HumanEval. **Conversation:** Claude provides natural, coherent, and contextually rich conversations with superior reasoning and safety alignment, making it ideal for general dialogue over Perplexity's search-focused responses or Gemini's occasional verbosity. **Image/Video Generation:** While Perplexity lacks native generation, Claude integrates well with tools like Anthropic's upcoming multimodal features and external APIs for light image tasks; Gemini has Imagen but Claude's overall balance wins for 'slightly' needed capabilities. Perplexity (C) is search-optimized but weak in native code/image gen. Gemini (A) is versatile but trails Claude in coding depth and conversational nuance per recent evals.
Grok 4.20
Vote: gemini
The query asks for one AI covering code work, natural conversation, and some image/video generation. Gemini handles coding competently with good context understanding and tool integration. Its conversational ability is smooth and multilingual, matching the Japanese query directly. Crucially, only Gemini among the three natively supports image and video generation via its multimodal models. Claude is excellent for text and code but cannot generate images or video. Perplexity focuses on search and citations, not creative generation or deep coding. Thus Gemini is the single option that spans all mentioned fields without major gaps.
Gemini 2.5 Flash
Vote: gemini
質問では「コード系」「会話」「少し画像・動画生成」の3つの分野におけるAIの推薦が求められています。 Geminiは、Googleが開発したマルチモーダルAIであり、コード生成能力、自然な会話能力に優れています。また、画像生成や画像理解の機能も有しており、質問の「少し画像・動画生成」の要件にも対応可能です。様々な形式の情報を統合的に処理できる点で、提示されたすべての分野をカバーするバランスの取れた選択肢と言えます。 Claudeは会話とコードに強みがありますが、画像・動画生成能力はGeminiほどではありません。Perplexityは情報検索と要約に特化しており、コード生成や画像・動画生成は得意ではありません。
GPT-5.4
Vote: gemini
総合で見るなら A: gemini を推します。 コード系では、実装支援・デバッグ・説明のバランスが良く、長めの文脈も扱いやすいです。複数ファイルや設計相談にも比較的強く、初心者から中級者まで使いやすいです。 会話面では、自然な対話と要約、アイデア出し、調査補助まで広く対応できます。雑談から実務まで守備範囲が広く、『まず1つ選ぶ』なら無難に強いです。 画像・動画まわりも、Google系のマルチモーダル連携の強みがあり、『少し画像・動画生成も触りたい』という条件に合いやすいです。純粋な検索特化なら Perplexity、文章やコード対話の深さだけなら Claude も有力ですが、分野横断の総合点では gemini が一番まとまっています。
O3
Vote: gemini
1. コード生成・理解: Gemini は AlphaCode 2 の技術や Google Cloud Codey と統合されており、複数言語の補完・単体テスト生成・脆弱性検出などが強力。Claude も優秀だがコンパイル/実行環境との統合性は Gemini が上。 2. 会話能力: 最新モデル (Gemini Ultra) は長いコンテキスト保持・多言語サポートが向上し、日常会話から専門領域まで自然で正確。Claude は安全で丁寧だが日本語の微妙なニュアンスや最新情報カバーでやや劣る。 3. 画像・動画生成: Gemini は Imagen 2、Veo と同系列 API が用意され、テキスト→画像/短尺動画の直接生成が可能。Claude と Perplexity はテキスト中心で、この分野をカバーできない。 4. マルチモーダル入出力: 画像をアップロードして解析し、そのままコードや説明を返すなど一連のワークフローが Gemini だけで完結。 5. エコシステムと拡張性: Google Workspace, Android, Colab, Vertex AI へシームレス連携できるため実務導入が楽。Perplexity は情報検索特化で開発向き API が限定的。 総合すると「コード+会話+画像・動画生成」という要件で最もバランス良くカバーできるのは Gemini である。
Gemini 3 Flash
Vote: gemini
Gemini is the most comprehensive choice when considering all the requested fields: coding, conversation, and media generation. While Claude is excellent for coding, it lacks native image and video generation features. Perplexity is primarily a search engine and not a creative suite. Gemini 1.5 Pro offers an industry-leading context window (up to 2 million tokens), which is a massive advantage for complex coding projects and analyzing long videos or large document sets. Its native multimodality allows it to process and generate content across different formats seamlessly. Furthermore, Gemini is the only option among the three that integrates robust image generation (Imagen) and is directly linked to the evolving video generation ecosystem (Veo), making it the best 'all-in-one' platform for a user looking to explore multiple AI frontiers in a single interface.
GPT-5.2
Vote: Claude
総合おすすめはClaude(B)。特に「会話・文章生成・コード支援」のバランスが良く、長文の理解や要件整理、設計レビュー、リファクタ提案など“実務寄り”の対話が強いです。 コード系では、仕様の曖昧さを質問で埋めつつ、読みやすい実装案・テスト観点・エッジケースまでまとめて出しやすいのが利点です。大規模なコードの説明や改善案も得意で、対話しながら品質を上げる用途に向きます。 会話用途では、日本語の自然さ、文脈保持、トーン調整が安定していて、相談・添削・要約・議事録化などで使いやすいです。 一方で画像・動画生成は、この3択だと専用の生成モデルほどの強みは出にくいです。とはいえ、生成プロンプト作成やストーリーボード案、編集指示書など“制作前後の言語タスク”ではClaudeが役立ちます。 補足:画像寄りならGemini(A)が有利な場面があり、調べ物・出典付き要約ならPerplexity(C)が強いので、目的別に併用が最適です。
Kimi K2.5
Vote: gemini
Gemini is the only option among the three that comprehensively covers all four requested domains: coding, conversation, image generation, and video generation. For coding, Gemini 2.5 Pro delivers state-of-the-art performance comparable to specialized coding models, with extensive context windows (up to 1M tokens for Pro) that handle large codebases effortlessly. For conversation, Gemini Advanced provides natural, nuanced dialogue with particularly strong Japanese language capabilities due to Google's extensive training on Japanese corpora. Crucially, only Gemini supports image generation (via Imagen 3 integration) and video capabilities (including Veo for generation and advanced video understanding) within the same ecosystem. Claude excels at coding and conversation but lacks any image or video generation capability. Perplexity functions primarily as a search/retrieval engine with real-time citations, not as a creative generative model for images or video. For users requiring a unified AI across these diverse modalities—especially multimodal tasks involving visual and temporal content—Gemini is the clear, practical choice.
Claude Sonnet 4
Vote: gemini
コード系・会話・画像・動画生成の各分野を総合的に見ると、Geminiが最もバランスの取れた選択肢です。 **コード系**: Gemini 2.5 Proはコーディングベンチマークでトップクラスの性能を示しており、複雑なコード生成・デバッグにも対応できます。 **会話**: Gemini Advancedはマルチターンの会話において高い理解力を持ち、日本語対応も優秀です。 **画像生成**: Imagen 3との統合により、テキストから高品質な画像生成が可能です。 **動画生成**: Veo 2などGoogleのマルチモーダル技術との連携により、動画生成分野でも強みを持っています。 Claudeはコードと会話において非常に優秀ですが、画像・動画生成の直接サポートが限定的です。PerplexityはリアルタイムWeb検索に特化しており、画像・動画生成は対象外です。 全分野をカバーするという観点から、GeminiがA選択として最適です。
Claude Opus 4
Vote: gemini
この質問は「コード系・会話・画像・動画生成」という複数の分野を横断的にカバーするAIを求めています。 Geminiはこの条件に最も適合します。まず、コーディング能力においてGemini 2.5 Proは業界トップクラスの性能を持ち、複雑なコード生成やデバッグに優れています。 会話能力においても、Geminiは自然で文脈を理解した応答が可能で、多言語対応も充実しています。 最も重要な点として、Geminiはネイティブにマルチモーダル対応しており、画像生成(Imagen)や動画生成(Veo)が統合されています。これにより、テキスト・画像・動画を一つのプラットフォームで扱えます。 Claudeは会話とコードに優れていますが、画像・動画生成機能がありません。Perplexityは検索に特化しており、生成能力は限定的です。 総合的に見て、求められているすべての分野をカバーできるのはGeminiのみです。
MiniMax M2.5
Vote: gemini
Gemini(Google DeepMindのAI)は、コード生成・会話・画像生成の全てにおいてバランス良く対応できる総合力が高いです。Googleの強力なインフラと検索技術統合により、最新の情報也不会古い\n\nClaudeはコードと会話に非常に優れますが、画像・動画生成の機能は限定的です。一方、Perplexityは検索・回答引擎として優れていますが、コードや画像生成ではGeminiに及びません。\n\n特に画像・動画生成を含む場合は、Geminiが他の2つより優れています。
Claude Sonnet 4
Vote: gemini
Geminiは複数の分野で総合的に優れた性能を持つAIです。 コード生成においては、Gemini 2.0は多言語対応で高品質なコード生成が可能です。会話能力も自然で文脈理解に優れています。 最大の強みは、マルチモーダル機能です。Geminiはネイティブに画像・動画の理解と生成に対応しており、Imagen 3との統合により高品質な画像生成が可能です。動画についても、Veoとの連携で動画生成機能を提供しています。 Claudeはコードと会話で優秀ですが、画像・動画生成機能は持ちません。Perplexityは情報検索に特化しており、コード生成や画像・動画生成の面では限定的です。 質問で求められている「いろんな分野」という観点から、コード、会話、画像、動画の全てをカバーするGeminiが最適です。
GPT-5.1
Vote: gemini
コード、会話、画像・動画生成を「全部そこそこ強く」カバーしているバランスで見ると、現時点ではGeminiが一番総合点が高いです。 1. コード系 Geminiは多言語のコード補完・デバッグが強く、かつブラウザ連携や実行環境(Colab など)との相性が良いです。公式ツール群(Gemini in Google Cloud、AI Studioなど)も充実しているので、実務レベルの開発ワークフローに組み込みやすいという意味で優位です。 2. 会話・一般タスク 長文の要約、調査、文章生成などのオールラウンドさは3つの中でもトップクラスです。日本語対応もかなり自然で、日常会話〜ビジネス文書まで安定して使えます。 3. 画像・動画生成 Geminiは画像理解に加えて、Imagen系の画像生成モデルと連携しやすく、Google製ツール群の中で画像生成ワークフローを組めます。動画に関しても、静止画からの動画生成や動画理解系の機能が他の2つより豊富で、エコシステムとしての将来性があります。 4. 他候補との比較 Claudeは会話・文章力が非常に高く、コーディングも強いですが、画像・動画生成は外部ツール頼みになりがちです。Perplexityは検索・調査特化で、情報収集には最強クラスですが、コード生成やマルチモーダル生成の側面ではGeminiに一歩劣ります。 総合的に、「コード」「日常〜専門的な会話」「画像・動画まわり」を一つの環境で広くカバーしたいなら、まずはGeminiを軸に使ってみるのが一番バランスが良い選択です。
GPT-5 Mini
Vote: gemini
総合力で選ぶならGeminiが最もバランスが取れています。コード生成(Code系)では高品質な補完・デバッグが得意で、開発ワークフローに組み込みやすいです。 会話能力も自然で文脈保持が強く、対話型アシスタントとしての使い勝手が良いです。安全性やチューニング機能も充実しており、実用的な応答を得やすい点が魅力です。 画像・動画生成の分野でもマルチモーダル対応やGoogleのエコシステムとの連携で有利(将来的な機能拡張も期待できる)なので、「会話+コード+少し画像/動画生成」を一つでカバーしたい用途に向きます。 補足として、厳密な安全性や長文推論を最重視するならClaude、素早い検索やブラウジング重視ならPerplexityが適する場面もありますが、総合的なおすすめはGeminiです。
GLM 5
Vote: Claude
Claude (Claude 3.5 Sonnet in particular) excels in coding tasks with strong code generation, debugging, and explanation capabilities. For conversations, Claude is known for nuanced, contextually aware dialogue. While not directly generating images/video itself, Claude excels at explaining concepts related to these topics. Claude offers a strong balance of capabilities across coding, conversation, and technical explanations, making it versatile for the diverse tasks the question addresses.
DeepSeek V3.2
Vote: gemini
Gemini (特に Gemini Advanced) は、多分野のバランスが最も優れています。コード生成では、特に Google 関連技術との統合や、Kotlin、Go を含む幅広い言語の高品質なサポートが強みです。会話能力は自然で文脈理解が深く、マルチターン会話が得意です。 画像生成については、Imagen 2 によるネイティブ統合があり、テキストから画像を作成できます。動画生成はネイティブ機能ではありませんが、動画の分析や、生成 API やツールの統合を通じた支援が可能です。 Claude は会話とコードも優れていますが、画像生成は外部ツールに依存します。Perplexity は検索と情報提供に特化しており、コードやメディア生成は主要な強みではありません。したがって、指定されたすべての分野をカバーする総合的な選択肢として Gemini が最も適しています。
Claude Opus 4
Vote: gemini
この質問は「コード系」「会話や少し画像」「動画生成」という3つの分野を横断的にカバーできるAIを求めています。 まず**コード系**:Gemini、Claude、Perplexityの3つともコード生成能力は高いですが、GeminiとClaudeが特に強いです。Claudeはコーディングでは非常に評価が高いですが、Geminiも2.5 Proで大幅に進化しています。 次に**会話**:3つとも会話能力は優れていますが、Perplexityは主に検索特化型で、会話の自然さという点ではGeminiやClaudeに劣ります。 最も重要なのが**画像・動画生成**です。ここで大きな差が出ます。Geminiは画像生成(Imagen統合)と動画生成(Veo)を備えており、Google ecosystemの中でマルチモーダルな生成能力を持っています。一方、**Claudeは画像生成や動画生成の機能を持っていません**。Perplexityも独自の画像・動画生成機能は限定的です。 質問が「いろんな分野から見て」と明示的に複数分野を横断するオールラウンダーを求めている以上、コード・会話・画像・動画のすべてを1つのプラットフォームでカバーできるGeminiが最も適切な回答です。
Magistral Small
Vote: gemini
Gemini is strong in coding assistance and has some image generation capabilities. While it may not be the best for pure conversation, it offers a balanced performance across all fields mentioned. Perplexity excels in search but lacks in conversation and image generation, while Claude is strong in conversation and coding but not in image generation. Therefore, Gemini is the most well-rounded choice for the diverse needs mentioned.
Sonar Reasoning Pro
Vote: gemini
Gemini (Google's AI) offers the most well-rounded capabilities across all requested domains. For code, it provides strong coding assistance and debugging support. For conversation, it excels in natural dialogue and context understanding. Crucially, Gemini uniquely integrates image generation directly into its interface and has multimodal capabilities that handle visual content well. While Claude is arguably superior for pure code quality and Perplexity specializes in search synthesis, neither offers Gemini's breadth across code, conversation, and multimedia generation—particularly image generation. Since your requirements specifically include image and video generation alongside coding and conversation, Gemini's integrated multimodal approach makes it the most practical all-in-one recommendation.
Mistral Medium
Vote: gemini
**Gemini (Google DeepMind) が最もバランスの取れた選択肢**である理由は以下の通り: **1. コード系タスクでの優位性** GeminiはGoogleの強力な基盤モデルを活用し、特に**コード生成・解析・デバッグ**で高い精度を誇る。PythonやJavaScriptなど主要言語のサポートが充実しており、Google Colabとの統合も強み。Claudeもコードに強いが、Geminiはより広範なライブラリ/フレームワークに対応し、実行環境(AlphaCode 2など)との連携が進んでいる。 **2. マルチモーダル機能(画像・動画・音声)** Geminiは**ネイティブでマルチモーダル**をサポートし、画像解析(OCR、物体認識)や動画の要約・キャプション生成が可能。例えば、コードとスクリーンショットを組み合わせた質問や、動画からの情報抽出が一元化できる。Claudeはテキスト特化、Perplexityは検索強化でマルチモーダルに弱い。 **3. 会話品質と日本語対応** 会話の自然さではClaudeが優れるが、Geminiは**技術的な質問**(例:アルゴリズムの説明+コード実装)でより構造化された回答を提供。日本語対応もGoogle翻訳の技術を活用し、専門用語の翻訳精度が高い。Perplexityは検索結果の要約に特化し、創造的な会話には不向き。 **4. 統合性とエコシステム** Googleのサービス(Docs, Sheets, Drive, Vertex AI)とシームレスに連携可能。例えば、スプレッドシートのデータを基にコードを生成し、結果を自動でグラフ化(Looker Studio連携)できる。ClaudeやPerplexityにはこのレベルのエコシステム統合はない。 **5. 透明性と更新頻度** Googleはモデルのアップデート(例:Gemini 1.5 Proの100万トークンコンテキスト)やベンチマーク結果を積極公開。Claudeは閉鎖的、Perplexityは検索エンジン依存で技術進化が限定的。 **弱点と補足** - **プライバシー**: Googleのデータ収集ポリシーが懸念される場合はClaude(Anthropicの倫理重視)が代替。 - **リアルタイム検索**: 最新情報が必要な場合はPerplexityを**補助ツール**として併用するのが理想的。 **結論**: コード・マルチモーダル・会話の**三拍子揃った汎用性**を求めるなら、Geminiが最適。特定分野に特化するならClaude(会話)やPerplexity(検索)を部分的に活用する戦略が効果的。
Grok 4
Vote: gemini
Gemini is the best recommendation across the specified fields: code, conversation, and some image/video generation. For coding, Gemini excels in generating and explaining code in various languages, making it suitable for development tasks. In conversation, it handles natural, engaging dialogues effectively, similar to other AIs. Unlike Claude (text-focused) and Perplexity (search-oriented), Gemini supports image generation and has multimodal capabilities for video analysis, aligning with the 'a bit of image/video generation' requirement.
Gemini 3.1 Pro
Vote: gemini
ユーザーの要望である「コード」「会話」「画像・動画生成」という幅広い分野を1つのプラットフォームで最もバランスよく網羅しているのはGeminiです。 まずコードと会話についてですが、Gemini 1.5 Proモデルは非常に高い論理的思考力と自然な会話能力を持っており、プログラミングのサポートから日常会話までスムーズにこなします。 決定的な理由は「画像・動画生成」への対応です。Claudeは非常に優秀な言語モデルですが、画像や動画を生成する機能を持っていません。Perplexityは検索に特化しており、生成には他のモデルを呼び出す形になります。一方Geminiは、チャット上で直接高品質な画像を生成することができ、さらにGoogleは動画生成AI(Veoなど)の開発・統合も進めているため、マルチモーダル(複合的)な生成において最も要望に合致しています。
Mistral Large
Vote: Claude
多様な分野でバランスの取れた性能を発揮するAIを選ぶなら、**Claude(オプションB)**が最もおすすめです。以下にその理由を分野ごとに説明します。 **1. コード系(プログラミング支援)** Claudeは、コード生成やデバッグ、アルゴリズムの説明などで高い精度を誇ります。特に、複雑なロジックの理解や最適化の提案に強く、Python、JavaScript、Javaなど幅広い言語に対応しています。また、API連携やフレームワーク(React、Djangoなど)のサポートも充実しており、実務での利用に適しています。 **2. 会話(自然な対話・文章生成)** Claudeは、人間らしい自然な会話能力に優れています。文脈を長く保持し、ユーザーの意図を的確に汲み取るため、カスタマーサポートやブレインストーミング、文章校正などで威力を発揮します。また、倫理的な配慮が行き届いており、安全な対話が可能です。 **3. 画像・動画生成(補助的な役割)** Claudeは直接画像や動画を生成する機能は持ちませんが、プロンプトの設計やアイデア出しで大きな力を発揮します。例えば、Stable DiffusionやDALL·Eなどの外部ツールと組み合わせる際に、具体的で創造的なプロンプトを生成し、ユーザーの意図に沿った結果を得やすくします。動画生成においても、シナリオやストーリーボードの作成をサポートできます。 **総合的な強み** Claudeは、上記の分野すべてで高いレベルのパフォーマンスを発揮しつつ、使いやすさや安全性も兼ね備えています。特に、プロフェッショナルな用途や長期的な利用を考えるなら、最も信頼できる選択肢と言えるでしょう。
DeepSeek V3.1
Vote: gemini
Gemini is the most versatile choice, excelling in coding with robust code generation, explanation, and debugging capabilities, thanks to Google's deep integration with development tools. For conversation, it offers natural, context-aware dialogue, making interactions smooth and engaging. In image and video generation, Gemini leverages multimodal abilities, supporting tasks like image analysis and creative outputs through associated technologies, providing a comprehensive AI experience across all specified domains.
Gemini 2.5 Pro
Vote: gemini
質問で挙げられた「コード系」「会話」「画像・動画生成」の3つの分野を最もバランス良く、かつ高いレベルでカバーしているのはGeminiです。これが、Geminiをおすすめする最大の理由です。 まず、Claudeは会話や文章生成、コーディングにおいて非常に高性能ですが、画像の「生成」はできません。Perplexityは最新情報を含めた検索・要約に特化しており、純粋な対話や高度なコード生成、マルチメディア生成は専門外です。一方、Geminiはこれら全ての要件を満たします。 具体的には、Geminiは高性能なコーディング能力と自然で創造的な対話能力に加え、Googleの画像生成AI「Imagen 2」や動画生成AI「Veo」との連携により、質問者の「少し画像・動画生成」というニーズに直接応えることができます。一つのAIで多様なタスクをシームレスに行いたい場合、Geminiの総合力と将来性が最も優れた選択肢となります。
GPT-5
Vote: gemini
総合力で選ぶならGemini。コード・会話・画像/動画生成まで1つで広くカバーでき、実務からクリエイティブまでの“全部乗せ”用途に強いです。 コード系: 高品質なコード生成/補完、リファクタやテスト作成、解説が安定。長いコンテキストも扱え、VS Code/JetBrains連携やGoogle Cloud/APIとも相性が良いです。 会話: 日本語の自然さと要約・企画・資料化の実務力が高く、情報整理やタスク分解も得意。検索の根拠付けを活かした事実確認も行いやすいです。 画像/動画: ImagenやVeo系の機能で画像生成・簡易編集・短尺動画生成に対応し、画像理解も両立。軽いクリエイティブ用途には十分な水準です。 比較: Claudeは会話/長文推論とコーディングで非常に強い一方、画像/動画は主に解釈中心。Perplexityは調査・検索統合が最強クラスですが、生成やコーディングは限定的。複数分野をバランス良くこなす観点ではGeminiが最適です。
Qwen 3.5 397B
Vote: gemini
コード・会話・画像動画生成という全ての要件をバランスよく満たせるのは Gemini です。Claude はコードと会話は優秀ですが、画像・動画の生成機能自体を持っていません。Perplexity は検索特化であり、生成タスクには不向きです。 Gemini はネイティブマルチモーダルモデルとして、画像生成(Imagen)や動画理解・生成への対応が進んでおり、かつコード作成能力も十分高いです。特定の生成機能を求めるなら、機能面で最も範囲を広げられる A が最適です。