無料
（通常 ¥600）
App Store で開く
アプリ内課金あり
iPhone / iPad 対応
57.0MB
年齢4+
MyLLM - Local AI Agentはどんなアプリ？
MyLLM - Local AI Agentは、端末内で動作する複数のモデルを選択して対話できるAIアシスタントです。
ネット環境がない機内や海外でも、自分のデバイス内で処理を行うためスムーズな会話が続きます。独自のPCやサーバーとQRコードで連携させることも可能で、ウェブ検索や画像生成など多機能なツールを使いこなせます。記憶を蓄積する機能が含まれており、個人の好みを学習しながらより精度の高い回答を導き出します。
自分のデータを外部のサーバーへ送信したくない場面や、プライバシーを守りながらAIを活用したい時に向いています。高度な機能を備えたエージェントとして、日々のタスクを効率的にこなしたい方にも重宝します。
アノアプリによる紹介文です。デベロッパー公式の説明はApp Storeでご確認いただけます。
MyLLM - Local AI Agentの注目ポイント
電波のない場所での対話
機内や海外など通信環境がない場所でも会話が途切れません。端末内で処理を行うため、データ通信を気にせずやり取りを続けられます。
複数のオープンモデルを選択
LlamaやQwenといった複数のモデルを本体で動かせます。自分の好みに合うものを選択して対話のスタイルを使い分けられる仕組みです。
外部機器との連携機能
QRコードを読み取るだけで、個人のPCやサーバーと接続できます。自宅の強力なマシンと連携してより高度な処理を行うことができます。
多機能なエージェントモード
ウェブ検索や画像生成といった道具を使いこなせます。メールの作成や地図の確認など、複数の機能を組み合わせてタスクをこなせます。
価格履歴
現在は無料です。過去134日間に1回値下げされ、値上げはありませんでした。
アノアプリが独自に収集した価格履歴をもとに分析しています。
- 買い時判定
- 通常価格
- 直前の価格から
- 100%オフ
- 最頻価格
- 無料
- 記録上の最安値
- 無料
価格更新履歴
| 日付 | 価格 |
|---|---|
| 2026-07-05 | 無料 |
| 2026-06-21 | ¥600 |
ランキング推移
ランキングデータがありません。
ランキング圏内（100位以内）に入ると記録が開始されます。
ユーザーレビュー （当サイト投稿）
まだレビューがありません。最初のレビューを書いてみましょう。
アプリ更新履歴 活発に更新中
Ver 6.4 2026-10-30
See the thinking
• Models that think before they answer show their thinking as it arrives. When the answer starts it folds to "Thought for 14s"; tap to read it again. Turn it off in Settings → Appearance.
Faster first words
• Replies from your own server or a cloud model now appear as they're written, even when the assistant has tools to hand.
Long chats with cloud models
• Cloud models now get the whole conversation their limit allows, so a long document stays in the chat for your follow-up questions. MyLLM uses each model's own limit; for a model it doesn't know, set it under Settings → Model → Context window.
Stay in charge of tools
• When a tool asks first, choose "Not now, and say why…" and the assistant is told your reason.
• Settings → Built-in Tools → Tool Permissions → Rules blocks a tool for good, or only when a call mentions something you name.
• When the assistant hands part of a job to a helper, a card shows what it's doing, with a Stop button.
Apps that check themselves
• Apps the assistant makes are opened out of sight first. If one hits an error, the assistant fixes it and the working app replaces the broken one.
Flashlight, camera and clearer app pages
• Gallery apps can use the flashlight, and the camera if you allow it. Try Signal Kit: Morse code by flashlight, screen or beep, and read it back.
• Each app's gallery page now says what it uses, such as the microphone or your location.
Work on Code
• Type @ and part of a file name to send that file with your message. No AGENTS.md yet? Project Rules → Write AGENTS.md for Me writes one.
Thank you for the reviews and suggestions — keep them coming.
• Models that think before they answer show their thinking as it arrives. When the answer starts it folds to "Thought for 14s"; tap to read it again. Turn it off in Settings → Appearance.
Faster first words
• Replies from your own server or a cloud model now appear as they're written, even when the assistant has tools to hand.
Long chats with cloud models
• Cloud models now get the whole conversation their limit allows, so a long document stays in the chat for your follow-up questions. MyLLM uses each model's own limit; for a model it doesn't know, set it under Settings → Model → Context window.
Stay in charge of tools
• When a tool asks first, choose "Not now, and say why…" and the assistant is told your reason.
• Settings → Built-in Tools → Tool Permissions → Rules blocks a tool for good, or only when a call mentions something you name.
• When the assistant hands part of a job to a helper, a card shows what it's doing, with a Stop button.
Apps that check themselves
• Apps the assistant makes are opened out of sight first. If one hits an error, the assistant fixes it and the working app replaces the broken one.
Flashlight, camera and clearer app pages
• Gallery apps can use the flashlight, and the camera if you allow it. Try Signal Kit: Morse code by flashlight, screen or beep, and read it back.
• Each app's gallery page now says what it uses, such as the microphone or your location.
Work on Code
• Type @ and part of a file name to send that file with your message. No AGENTS.md yet? Project Rules → Write AGENTS.md for Me writes one.
Thank you for the reviews and suggestions — keep them coming.
Ver 6.3 2026-10-28
Work on Code
• The + button → Work on Code opens a folder of code (in Finder on a Mac, in Files on iPad and iPhone). Ask for a fix or a feature and the assistant finds the right files, reads them and makes the change.
• Every change is saved into the folder and listed with a diff. Undo the last change, revert a file or everything, or keep it all.
• Add an AGENTS.md file to give the assistant your project's rules. MyLLM tells you which commands or tests to run. Needs your own server or a cloud model.
Connectors
• Settings → Connectors: sign in to services such as Notion, Linear, Jira, Figma, Dropbox and Supabase, and the assistant can use them in your chats.
• Your sign-in stays on this device and requests go straight to the service. Tools that change something ask you first.
Check This Answer
• Long-press a reply → Check This Answer for a second opinion from Gemma 4 on your device, with its working.
Also: OpenCode Go and MiniMax Coding Plan subscriptions, and an updated privacy policy.
Thank you for the reviews and suggestions — keep them coming.
• The + button → Work on Code opens a folder of code (in Finder on a Mac, in Files on iPad and iPhone). Ask for a fix or a feature and the assistant finds the right files, reads them and makes the change.
• Every change is saved into the folder and listed with a diff. Undo the last change, revert a file or everything, or keep it all.
• Add an AGENTS.md file to give the assistant your project's rules. MyLLM tells you which commands or tests to run. Needs your own server or a cloud model.
Connectors
• Settings → Connectors: sign in to services such as Notion, Linear, Jira, Figma, Dropbox and Supabase, and the assistant can use them in your chats.
• Your sign-in stays on this device and requests go straight to the service. Tools that change something ask you first.
Check This Answer
• Long-press a reply → Check This Answer for a second opinion from Gemma 4 on your device, with its working.
Also: OpenCode Go and MiniMax Coding Plan subscriptions, and an updated privacy policy.
Thank you for the reviews and suggestions — keep them coming.
Ver 6.2.1 2026-10-24
More subscriptions
• OpenCode Go: use your OpenCode Go subscription for GLM, Kimi, DeepSeek, Qwen, MiniMax and more. Settings → Connection → Cloud → OpenCode Go, then paste your key.
• MiniMax: use a MiniMax Coding Plan key or a normal MiniMax key. Its reasoning stays out of the reply, so answers come through clean.
Thank you for the reviews and suggestions — keep them coming.
• OpenCode Go: use your OpenCode Go subscription for GLM, Kimi, DeepSeek, Qwen, MiniMax and more. Settings → Connection → Cloud → OpenCode Go, then paste your key.
• MiniMax: use a MiniMax Coding Plan key or a normal MiniMax key. Its reasoning stays out of the reply, so answers come through clean.
Thank you for the reviews and suggestions — keep them coming.
Ver 6.2 2026-10-23
MyLLM 6.2
RICHER REPLIES
• Longer answers from your own server or a cloud model now read like a well-designed page: the answer first, then clear sections, key words in bold, steps, tables, and features or pros and cons laid out with an icon beside each one.
• Replies sit on the page at full width, like an article, and are formatted while they arrive instead of appearing as plain text until the end.
• Code is coloured by language, with a Copy button on every block. Maths is drawn properly: fractions, roots and powers.
• Photos fill the width with a label saying where they came from, and cited sources show as small labels you can tap.
• Lists can have sub-points, and checklists show ticks.
A ROOMIER MESSAGE BOX
• The message box runs the full width of the screen, with send inside it. Attachments and Tool Focus are under the + on the left.
FIXES
• Tables now show all their columns. Before, only the first one was visible.
• Long lines of code scroll sideways instead of being cut off at the edge.
Prefer the old look? Settings → Appearance & Interaction → Rich reply layout.
RICHER REPLIES
• Longer answers from your own server or a cloud model now read like a well-designed page: the answer first, then clear sections, key words in bold, steps, tables, and features or pros and cons laid out with an icon beside each one.
• Replies sit on the page at full width, like an article, and are formatted while they arrive instead of appearing as plain text until the end.
• Code is coloured by language, with a Copy button on every block. Maths is drawn properly: fractions, roots and powers.
• Photos fill the width with a label saying where they came from, and cited sources show as small labels you can tap.
• Lists can have sub-points, and checklists show ticks.
A ROOMIER MESSAGE BOX
• The message box runs the full width of the screen, with send inside it. Attachments and Tool Focus are under the + on the left.
FIXES
• Tables now show all their columns. Before, only the first one was visible.
• Long lines of code scroll sideways instead of being cut off at the edge.
Prefer the old look? Settings → Appearance & Interaction → Rich reply layout.
Ver 6.1 2026-10-22
MyLLM 6.1
SHARED LISTS
• Shopping List and Quick Todo can now hold several lists. Tap + New to start another: one for the weekly shop, one for a party, one for the holiday.
• Share a list with family or friends through iCloud. Tap Share, send the invitation, and when they open it the list appears in their own MyLLM. Everyone's changes show up for everyone, even when two of you edit at the same moment.
• A list stays on your device until you share it. Whoever made a shared list can rename it, change who's in it or delete it for everyone; anyone else can leave it.
• Ask the assistant to add to a list by name, like "add balloons to my party list".
• Works on iPhone, iPad and Mac. Everyone needs MyLLM 6.1 and an iCloud account.
GALLERY
• The app gallery now loads from MyLLM's own gallery server, which checks it is talking to the genuine MyLLM app using Apple's App Attest. Nothing personal is sent. If the server can't be reached, the gallery falls back to its public copies as before.
SHARED LISTS
• Shopping List and Quick Todo can now hold several lists. Tap + New to start another: one for the weekly shop, one for a party, one for the holiday.
• Share a list with family or friends through iCloud. Tap Share, send the invitation, and when they open it the list appears in their own MyLLM. Everyone's changes show up for everyone, even when two of you edit at the same moment.
• A list stays on your device until you share it. Whoever made a shared list can rename it, change who's in it or delete it for everyone; anyone else can leave it.
• Ask the assistant to add to a list by name, like "add balloons to my party list".
• Works on iPhone, iPad and Mac. Everyone needs MyLLM 6.1 and an iCloud account.
GALLERY
• The app gallery now loads from MyLLM's own gallery server, which checks it is talking to the genuine MyLLM app using Apple's App Attest. Nothing personal is sent. If the server can't be reached, the gallery falls back to its public copies as before.
Ver 6.0 2026-10-19
MyLLM 6.0
NEW
• Agent connection: a coding agent on your computer can now build and test your MyLLMos apps on this device. It works with any app or tool that speaks MCP. Turn it on in Settings → Agent connection and paste the setup it gives you into your agent.
• The agent can read and change your apps, run them on this device and read the errors, open one and take a screenshot, and try prompts on this device's own model to see how fast it is.
• MyLLM asks you first before an agent changes an app, writes a file, opens an app or looks at your screen. You can allow that for 30 minutes at a time. It's off until you turn it on, and every request needs the key Settings shows.
• Work on a document: open a text, PDF or Word file beside the chat. The assistant proposes edits as tracked changes you accept or reject one by one. Text files are saved in place; PDF and Word edits are saved as a copy. You can also choose a folder of your documents, or drop a file onto the chat on a Mac or iPad.
• A bigger message box: it now grows as you type, so you can read over a long or pasted message before sending. Tap the expand arrow to see the whole message on one screen. Return now starts a new line; tap the arrow to send.
NEW
• Agent connection: a coding agent on your computer can now build and test your MyLLMos apps on this device. It works with any app or tool that speaks MCP. Turn it on in Settings → Agent connection and paste the setup it gives you into your agent.
• The agent can read and change your apps, run them on this device and read the errors, open one and take a screenshot, and try prompts on this device's own model to see how fast it is.
• MyLLM asks you first before an agent changes an app, writes a file, opens an app or looks at your screen. You can allow that for 30 minutes at a time. It's off until you turn it on, and every request needs the key Settings shows.
• Work on a document: open a text, PDF or Word file beside the chat. The assistant proposes edits as tracked changes you accept or reject one by one. Text files are saved in place; PDF and Word edits are saved as a copy. You can also choose a folder of your documents, or drop a file onto the chat on a Mac or iPad.
• A bigger message box: it now grows as you type, so you can read over a long or pasted message before sending. Tap the expand arrow to see the whole message on one screen. Return now starts a new line; tap the arrow to send.
Ver 5.9.8 2026-10-16
MyLLM 5.9.8
FIXES
• Own server (Ollama): chats no longer lose track of earlier messages. A model MyLLM didn't know by name was treated as having a small memory, so earlier messages were left out; it now uses the context size you set.
• Saying "proceed" or "yes" to a question the assistant asked works again: your previous message and its reply are always sent.
• The assistant can now try up to three rounds of tools before answering, so "let me try something else" is followed through.
FIXES
• Own server (Ollama): chats no longer lose track of earlier messages. A model MyLLM didn't know by name was treated as having a small memory, so earlier messages were left out; it now uses the context size you set.
• Saying "proceed" or "yes" to a question the assistant asked works again: your previous message and its reply are always sent.
• The assistant can now try up to three rounds of tools before answering, so "let me try something else" is followed through.
Ver 5.9.7 2026-10-14
MyLLM 5.9.7
NEW
• Edit your apps with the model on this iPhone: Shell now shows your saved apps as files in its apps/ folder. Ask MyLLM there to change one and it reads only the lines it needs, shows you each edit before it's made, and checks the app for errors afterwards. Works with no signal.
• Only Shell can open your saved apps, and each change keeps the version it replaced, so undo puts it back.
FIXES
• Edit in Chat now asks first when your chats use the model on this iPhone, and offers to open the app in Shell or switch to your own server or a cloud model. Before, the phone model was asked anyway and could only say it didn't have the tools.
• A request from Edit in Chat or Remix is no longer sent until you've typed the change you want.
NEW
• Edit your apps with the model on this iPhone: Shell now shows your saved apps as files in its apps/ folder. Ask MyLLM there to change one and it reads only the lines it needs, shows you each edit before it's made, and checks the app for errors afterwards. Works with no signal.
• Only Shell can open your saved apps, and each change keeps the version it replaced, so undo puts it back.
FIXES
• Edit in Chat now asks first when your chats use the model on this iPhone, and offers to open the app in Shell or switch to your own server or a cloud model. Before, the phone model was asked anyway and could only say it didn't have the tools.
• A request from Edit in Chat or Remix is no longer sent until you've typed the change you want.
Ver 5.9.6 2026-10-13
MyLLM 5.9.6
FIXES
• Long tasks on your own server (like editing a saved app) no longer get cut off part-way. MyLLM now makes sure every step leaves room for the model's reply.
• If your server adds its own text to each request (a proxy adding search results, say), MyLLM now reads how big the request really was and leaves room for that too.
• If a step is still cut off, it isn't run, and the assistant tells you that change wasn't made.
• If a conversation is too full to go on, the assistant stops and says so. Start a new chat to carry on.
• Big changes to saved apps are made in smaller steps, so they fit.
FIXES
• Long tasks on your own server (like editing a saved app) no longer get cut off part-way. MyLLM now makes sure every step leaves room for the model's reply.
• If your server adds its own text to each request (a proxy adding search results, say), MyLLM now reads how big the request really was and leaves room for that too.
• If a step is still cut off, it isn't run, and the assistant tells you that change wasn't made.
• If a conversation is too full to go on, the assistant stops and says so. Start a new chat to carry on.
• Big changes to saved apps are made in smaller steps, so they fit.
Ver 5.9.5 2026-10-12
MyLLM 5.9.5
NEW: Phone helper (Settings → Privacy)
• Personal data stays on this phone: when you chat with a cloud model or your own server, your files, email, contacts, calendar, photos and other personal data are read by the model on your iPhone, and only its answer is sent.
• Condense long pages on this phone: long web pages are summarised on your iPhone first, so the cloud model reads less and costs less.
Both are off until you turn them on.
FIXES
• Browse Hugging Face now checks each build against the memory iOS actually gives MyLLM on your iPhone, so it no longer suggests a model that downloads but can't load.
• If a model is too big to load, the message now says how much memory it needs and how much is available, and suggests a smaller build.
• The assistant is no longer told about tools it doesn't have in that turn.
NEW: Phone helper (Settings → Privacy)
• Personal data stays on this phone: when you chat with a cloud model or your own server, your files, email, contacts, calendar, photos and other personal data are read by the model on your iPhone, and only its answer is sent.
• Condense long pages on this phone: long web pages are summarised on your iPhone first, so the cloud model reads less and costs less.
Both are off until you turn them on.
FIXES
• Browse Hugging Face now checks each build against the memory iOS actually gives MyLLM on your iPhone, so it no longer suggests a model that downloads but can't load.
• If a model is too big to load, the message now says how much memory it needs and how much is available, and suggests a smaller build.
• The assistant is no longer told about tools it doesn't have in that turn.
