Changelog.

📐 Diagrams That Hold Together

• Tables and text diagrams finish drawing. The safety check that watches for stuck output used to occasionally cut off the neat, repetitive-looking parts of an answer (table rows, the lines and boxes of a text diagram). Those now get much more room, while genuinely stuck output is still caught right away.
• Diagrams export faithfully. Flow diagrams now export to HTML and PDF looking exactly as they appear in the app, instead of sometimes losing their labels.
• Captions that carry. Diagram captions show up consistently under the diagram, both in the app and in exported files.

🎨 A Calmer, Cleaner Canvas

• Smoother to move things. Dragging, resizing, and panning now track your cursor immediately, without the slight lag that made precise placement feel imprecise.
• A toolbar that only shows what applies. Color swatches and an incomplete "connect" button that did nothing on certain items have been removed, so the selection toolbar only offers actions that actually work on what you've selected.
• Connections work as before. You still connect ideas by dragging from one item to another.

🛠️ Smaller Fixes & Refinements

• Export menu stays on top. The Export dropdown on a research note no longer gets clipped behind the bar beneath it, so every option is reachable.
• Cleaner Evidence Ledger. Each claim now shows its text without stray citation markers mixed in, and the progress bar counts up smoothly without appearing to overshoot.
• Truer to the chat. A handful of small rendering touch-ups mean a saved research note lines up more closely with what you watched stream into the chat.

📚 Research notes that look published

When you save a research answer, you now get a polished document instead of a wall of text.

• A real title cover with the date and a live count of how many sources informed the work.
• Native charts — genuine bar, line, pie, and scatter charts with axis labels, value labels, gridlines, and legends, able to pull their data straight from a table or the evidence ledger.
• Callout boxes — highlight key points in tidy info, insight, warning, or success callouts.
• Captioned diagrams that render reliably, with stray or mis-typed diagrams quietly repaired.
• One clean reference list — duplicate bibliographies are stripped, and the evidence ledger tracks every claim.

Export that matches what you see

Export any research note to HTML, PDF, or Markdown and the file looks exactly like the one on your screen — diagrams, charts, math, tags, and table of contents included. Exports are theme-aware, so a light vault produces a light document and a dark vault a dark one.

⚡ Turbo — faster local AI

An optional new engine runs supported local models up to about twice as fast, with no change in answer quality. Standard stays the default.

• One tap to switch — supported models show a purple "⚡ Turbo" pill.
• Live speed readout — watch real-time tokens-per-second, prefill progress, and thinking state in the sidebar.
• Clear trade-offs — Turbo is a separate download and is text-only, so image input switches off while it's active; the app spells this out before you commit.
• Manage downloads — a new dialog lets you delete the Standard build, the Turbo build, or both.
• Smoother long answers — streaming keeps up with big drafts instead of turning choppy.

💭 Watch the model think

A model's private reasoning now streams into a quiet, collapsible thought card instead of leaking into the answer. You can watch it think, then it tucks itself away once the real answer begins — and that reasoning never ends up in your copies, exports, or follow-up turns.

🖋️ Richer, calmer chat

• Diagrams that draw as they stream, holding the last good frame instead of flickering.
• Collapsible toggles and callout boxes in replies, with smooth open/close animations.
• Crisper math, and pasting a table full of dollar signs no longer gets mistaken for an equation.
• Expressive when it helps, concise when it should be — replies match the depth of your question.
• Gentle motion — new messages slide in and the view follows along, all respecting your reduced-motion setting.

🔎 Smarter, fairer research

• Hears both sides — on genuinely debatable questions, Canto seeks out opposing evidence and files it in the Evidence Ledger's "Against" column. Neutral topics are left alone, so there's never manufactured controversy. A simple Auto / On / Off control puts you in charge.
• "Expand on [W101]" — refer to a source by its citation marker and Canto re-reads exactly that source instead of guessing with a new search.
• Clearer save feedback — a real progress bar counts claims as they're extracted, and the "Created…" confirmation is a link that opens your new note.
• Save more than once — turn a single rich chat into several documents; the save bar stays ready with a "Save another as" option.

🕸️ Smarter connections everywhere

• Consistent suggestions — the related-notes list, the knowledge graph, and chat grounding now all rank notes the same way, so they stop contradicting each other.
• Short AI summaries of each note power cleaner, more relevant related-note matches.
• Natural graph navigation — scrolling now pans the knowledge graph, and Ctrl/Cmd+scroll (or a trackpad pinch) zooms toward your cursor.

🔌 A bigger door for connected tools

For people who link Canto to external AI clients, connected tools can now read note details and chat transcripts and drive more of the app (chat, sidebar, Vault Manager, Settings, and graph) — all still on your terms.

🛠️ Fixes & refinements

• Local Ollama models now report the context size they actually loaded, so what you see always matches reality.
• On notes with a cover and a long table of contents, the tag row no longer overlaps the contents.
• Clicking into a note's body below a collapsed diagram or the tag row now places your cursor correctly.
• A safety cap prevents weaker models from hanging on the faster Turbo engine.
• Collapsible sections export to Markdown respecting whether they were open or closed.

🎛️ The Right Settings in the Right Place

• Context-window controls moved to Settings → Agent, right next to auto-compaction, so everything about your conversations lives together.
• Choose Adaptive (Canto sizes the context window to your machine) or Manual (pick the exact size yourself), with plain-language descriptions for every option.
• A clear Reload model button applies your changes immediately.
• A brand-new Settings → Advanced panel gathers the deeper, optional knobs — Low Memory Mode, GPU/CPU balance, and idle-model timers — into one clearly labelled place, with a "Reset to defaults" link one click away.
• Behind the scenes, Canto now tidies up local models it loads while checking them, so they don't quietly pile up in memory on smaller machines.

📚 Polished Research, Everywhere

• The Document Agent now produces research notes to the same high standard as Canto Chat.
• Numbered citations in agent-written research turn into tidy, clickable citation pills.
• Sources gather into a single, clean References section instead of scattered or duplicated lists.
• The full bibliographic detail a model writes for a source is preserved, so references read properly rather than getting trimmed to a bare link.
• Diagrams in your notes can now carry a short caption underneath, so it's clear what each one shows.

🏷️ Chat Titles That Actually Fit

• Auto-generated chat titles now reflect what the assistant actually answered, not just your opening question.
• The model's behind-the-scenes "thinking" never leaks into a title anymore.
• Opening Agent Chat no longer pops the slash-command menu open on its own — it waits until you type "/".

🎨 A Calmer, Cleaner Window

• The sidebar now reads as its own gentle "island" while the editor, tabs, and status bar settle onto a soft, neutral page surface.
• The sidebar smoothly crossfades as you switch between Canto Chat, the Welcome screen, and the editor, rather than snapping.
• The cleaner surfaces look great across all of Canto's light and dark themes.

🛠️ Smaller Fixes & Refinements

• Update notification pills are easier to click precisely, and the Restart button can't be triggered twice while an install is already underway.
• While a new version is installing, the button shows a spinner and an "Installing" label, then recovers gracefully if anything goes wrong.
• Canto recognizes your existing References, Sources, and Footnotes sections more reliably when adding to a note.
• Code blocks and their toolbars now have crisp borders in light themes, matching the polish of dark mode.

💭 Watch your model think

• See the reasoning unfold — models that "think" before they answer now stream that step-by-step reasoning into Canto as it happens.
• Cleanly separated — the thinking is tucked into its own collapsible block, so your final answer stays clean and easy to read.

📏 Context sizes you can trust

• Real numbers, not a placeholder — local connections no longer show a misleading "256K". Canto discovers a model's actual limit where it can, and falls back to a sensible value otherwise.
• Set it exactly — a handy preset picker lets you tell Canto a model's true window in a couple of clicks, and the context meter updates instantly.
• The meter finally agrees — the "how full is the model" bar now reflects your real limit, so warnings and auto-compaction kick in at the right time.
• Your settings stick — a context size you set by hand is remembered and survives refreshing your list of models.

🛠️ Smarter tool detection

• A fair test — Canto now gently warms a model up before checking whether it can use tools (reading notes, searching the web, editing documents), so big models aren't penalized for being slow to load.
• More patience for heavy models — the check waits longer for large local models, so a capable 9B model is recognized just like a quick 4B one.
• Have the final say — if the automatic check ever gets it wrong, you can switch tool support on or off for any model yourself.

🎯 Friendlier limits & refinements

• Plain-language memory warnings — if a local model runs short on memory mid-request, Canto explains what happened and suggests how to fix it, instead of showing a cryptic error.
• Cleaner structured replies — oMLX models now return tidy, well-formed results when Canto asks for structured output.

♻️ Long chats that never hit a wall

• Auto-compaction. When a chat starts to fill up, Canto quietly summarizes the older turns into a short recap and keeps the recent ones word-for-word — so you can keep going without losing the thread.
• Recent turns stay exact. Your latest messages are always kept verbatim, never summarized.
• "Compact now" button. Reclaim space yourself anytime, right from the context meter.
• Tune it your way. New Agent settings let you choose how soon it kicks in, how many recent turns to always keep, who writes the recap, and whether to get a gentle heads-up when it happens.
• Works everywhere. Applies to Canto Chat, the Document Agent, and Vault Chat.

📊 Diagrams you can actually explore

• Pan, zoom, and go fullscreen. Get in close on a complex diagram or open it full-screen to study the whole thing.
• Captions. Diagrams can carry a short title so you know what you're looking at.
• One-click repair. If a diagram comes out broken, a "fix" action asks Canto to mend it — and you can watch it work.
• Edit it yourself. A split-view editor lets you tweak a diagram by hand.
• Smoother while streaming. Diagrams now settle into place cleanly as a reply comes in, instead of flickering or grabbing your scroll.

🛤️ Vault Manager shows its work

• Live timeline. Vault Manager now reports what it's doing step by step — matching the look of Canto Chat and the Agent Panel — instead of a single vague status line.
• Grouped results. Drafted changes are tallied together, and anything skipped (duplicates, limits, or errors) is shown plainly.
• Peek at its thinking. Expandable thought traces let you follow its reasoning when you want to.
• Nothing hidden. If a run stops early for any reason, it tells you why.

🎯 Calmer, steadier research notes

• One health indicator. Ledger problems, uncitable claims, and weakly-supported claims now live in a single tidy popover next to the note title, instead of scattered banners.
• Citations that survive follow-ups. Ask a research note a follow-up question and its citation numbers keep resolving correctly across turns.
• Honest saves. Saving a research note always keeps the full draft and clearly flags anything that got trimmed or dropped, rather than quietly swapping in a shorter answer.
• Sturdier evidence ledger. The claims-and-sources table is more reliable to build, and claims trace back to the sources you actually see cited.

🛠️ Smaller fixes & refinements

• Roomier chat layout. The Canto Chat split and editor minimum width were retuned so narrow side-by-side layouts stay comfortable to work in.
• Clearer embedding status. The background indexing indicator now names the note or link it's currently working on.
• Protected note metadata. Research notes are better guarded against accidentally losing their title or misplacing the tag row while you edit.
• Tidier numbers and tables. Context-window sizes display as clean round values, and currency symbols in tables are no longer mistaken for math.

🔗 Citations that belong to the note

• References are now owned by the note itself, so numbering is always correct.
• Reorder your sources and the whole note renumbers itself to match.
• Citations stay linked to the right source even as you edit.
• The References section is protected so the AI can't accidentally scramble it.

📚 A reference registry you can manage

• A dedicated panel lists every source a note cites, with how often each one is actually used.
• Reorder, edit, and add sources — the note updates to match.
• "Remove all uncited" clears out sources that were gathered but never used.
• See when a reference is also used in other notes, so you don't lose useful ones.
• "Rebuild from body" reconstructs the registry from the note's content if things ever drift.

🔎 Citation hover cards

• Hover any citation to see the source, domain, and a quick reliability read — without leaving your place.
• See how often the source is cited here and whether it appears in other notes.
• Jump to the source or work with the reference right from the hover card.

⚠️ Citation health check

• When Canto writes a research note, it tells you upfront how many claims it could and couldn't back with a source.
• Citation markers that don't point to a real source are cleaned up instead of left dangling.
• Dismiss the banner anytime, or tell it not to show again for that note.

🧾 Evidence Ledger

• A structured table of a note's key claims, the sources supporting (or contradicting) each, and a confidence rating.
• Color-coded high / medium / low confidence at a glance.
• Thinly-supported or contradicted claims get a subtle underline in the note body.
• "Insert as section" makes the ledger a permanent part of the document when you want it.

🔄 Living research notes

• Come back to a Canto-authored research note later and choose "Update this research" to re-run it.
• A merge view summarizes new sources, new or revised claims, confidence shifts, and new contradictions before anything changes.
• Review the diff and choose to merge or discard — a refresh never silently rewrites your work.

🎯 A calmer, faster vault

• Related Notes suggestions are sharper and arrive with less churn.
• The Knowledge Graph stays responsive on very large vaults and tells you when it's trimmed for speed.
• An open Knowledge Graph tab now restores after you restart Canto.
• The local AI's indexing engine no longer gets unloaded mid-job, so background work finishes cleanly.

🛠️ Smaller fixes & refinements

• Stray internal scaffolding (like "next steps") no longer leaks into notes Canto writes for you.
• Saving a research note never drops a citation it had just shown you.
• A labeled spinner shows when Canto is building citations, the ledger, and references.
• Research notes get tidy role labels in the margin (Summary, Methodology, References) to make long documents easier to scan.

👀 Know the Impact Before You Confirm

• Before deleting, Canto shows you how many notes will be affected and which ones.
• You can see how much indexed content will be removed, with a clear warning that the action can't be undone.

📎 Note Attachments Join the Library

• Files attached directly inside notes now appear in the Library, marked as Indexed.
• These attachments are included in vault-wide search when Canto Chat is set to use all your notes.
• Whether a file came from chat or from a note, it now lives in the same Library view.

🪄 Smarter Drag-and-Drop

• Dragging a file from the sidebar into the editor now drops a real attachment you can open — not just plain text.
• Dragging a note in creates a working link instead of the note's title as text.
• The Library refreshes quietly in the background and attachment indicators update on their own.

🧰 A Better Vault Manager

• The Vault Manager chat now uses the same polished composer as Canto Chat, with quick-action presets built in.
• You can choose your model directly from the Vault Manager.
• A context meter shows how much room the vault scan, note reads, and the drafted plan are using as the agent works.
• Smarter handling avoids repeating the same folder actions on smaller models.

📊 Live Context Usage

• The context meter now updates while Canto is working, not just when it finishes — so you can see context fill up in real time.

🧠 Memory, One Click Away

• Open the adaptive Memory dashboard right from Settings → AI Models.
• High-memory Macs now get higher recommended context limits, so guidance matches what your machine can handle.

🧺 Inbox Mode Polish

• Related-note suggestions are now drawn from real source content, not just titles and metadata.
• A compact run strip stays pinned near the composer, so you can apply, retry, or jump to a run without scrolling.
• A new Start Over option discards a suggested plan and returns you to the composer — without touching your vault.
• Citations stay cleaner: existing labels are preserved and reference lists stay tidy.

🛠️ Smaller Fixes & Refinements

• Attachment indicators update right after a file is recognized — no manual reload needed.
• The Library refreshes while open without flickering between loading and error states.
• Files mistakenly marked as failed to index can now recover on their own.
• Stray internal text no longer slips into notes Canto creates for you.

Notes, Notebooks, and Organization

• Notes and notebooks created from chat can now be placed in a chosen folder.
• Approval previews now show the destination folder before a new note or notebook is created.
• Source links and reference sections are more consistent when Canto creates or updates documents.

Memory Links and Search

• Improved how Canto prepares your vault for Memory Links, Related Notes, and source matching.
• Large vaults should catch up faster in the background.
• Manual Memory Links rebuilds are more dependable.
• Memory settings now better reflect what has actually been prepared.
• Repeated source links are handled more calmly, reducing duplicate background work.

Library and Files

• Library items now show clearer indexing/preparation progress.
• Failed Library preparation can be retried more easily.
• Image-aware chat behaves more steadily while visual context is being prepared.
• Asset indexing progress is easier to see from the Library.

Voice

• Added Italian support for transcription.
• Improved the voice model list with clearer recommendations for higher-quality transcription.
• Made voice model choices easier to compare and manage.

Fixes and Polish

• Reduced duplicate work when the same link is used more than once.
• Improved source handling for Inbox Mode and created notes.
• Improved manual rebuild behavior for Memory Links.
• Made follow-up research feel more responsive.
• Added general polish around chat, Library preparation, Memory Links, and voice settings.