Tutorials: Anki 25.09 Deep Dive: Per-Deck FSRS Customization, Better Analytics, and the New Launcher Architecture

.

Anki 25.09 Deep Dive: Per-Deck FSRS Customization, Better Analytics, and the New Launcher Architecture

 

The open-source flashcard ecosystem has witnessed a massive evolutionary leap since the introduction of the Free Spaced Repetition Scheduler (FSRS). Moving away from the legacy SM-2 algorithm, FSRS introduced neural-network-backed scheduling that radically optimizes memory retention while minimizing review fatigue.

With the launch of Anki 25.09, the development team at Ankitects has rolled out structural overhauls to both backend mechanics and core installation architecture. This release refines how algorithmic scheduling handles decentralized data, upgrades workload simulators, simplifies platform navigation via an automated native launcher, and patches critical zero-day vulnerabilities.

1. Algorithmic Upgrades: The New FSRS Paradigm

The headline feature of Anki 25.09 is the profound decoupling of mathematical parameters from rigid global presets. In previous versions, tweaking your Desired Retention (DR) meant duplicating options profiles and assigning separate presets to each distinct branch of your deck topology. Anki 25.09 thoroughly re-architects this relationship.

Per-Deck Desired Retention (DR)

You can now natively define isolated Desired Retention metrics down to the individual deck tier without generating independent, messy option profiles. This allows for a streamlined paradigm:

• Highly volatile or complex information (such as medical pathology or advanced programming syntax) can be configured to a high target retention rate (e.g., 90%–92%).

• Low-stakes vocabulary or auxiliary reading sets can concurrently target an optimized 75%–80% retention profile inside the exact same preset group.

The underlying engine recalculates states seamlessly on the fly. Whenever you update a subdeck's specific desired retention parameter, the background database triggers a real-time computation to re-evaluate the memory state and adjust relative intervals across those nodes instantly.

The Shift to "Help Me Decide" (Visual Workload Simulation)

Historically, adjusting target retention felt like shooting in the dark, mediated by the complex Compute Minimum Recommended Retention (CMRR) framework. Anki 25.09 deprecates CMRR in favor of an interactive, data-driven "Help Me Decide" mini-simulator interface.

When you click "Help Me Decide" within the deck configuration panel, Anki maps a dynamic Time / Memorized Ratio graph. Instead of letting a hidden mathematical equation guess your threshold, the software simulates how your card load scales relative to target retrievabilities over a defined chronological horizon.

   Review Time / Workload
     ^
     │          / (95% DR: High effort, low interval)
     │         /
     │        /
     │  ─────┘ (Sweet Spot)
     │ / 
     │/  (70% DR: Low effort, poor recall)
     └───────────────────────────────> Desired Retention %

The mini-simulator balances computational performance with accuracy, delivering calculations within a 25% margin of error compared to a full, multi-minute deck simulation. This tool provides visual clarity on exactly where your review workload spikes exponentially, allowing you to establish a sustainable middle ground.

2. Infrastructure Shift: The Standalone Anki Launcher Framework

Deploying and maintaining Python/Qt-based applications across fragmented operating systems has long been a logistical bottleneck for open-source software. Anki 25.09 addresses this by formalizing an entirely new native launcher framework as the default vector for desktop deployment across Windows 10/11, macOS 12+, and Linux.

Dynamic Upgrades/Downgrades

Rather than forcing you to completely purge file directories, download bulky external installers, and risk system registry contamination, the desktop container functions via a modular environment. Under Tools -> Upgrade/Downgrade, the core application binaries can be dynamically switched, updated, or re-vectored from within the UI.

Sandboxed and Portable Routing

The new system introduces a native mechanism allowing users to instruct the launcher to route assets and metadata to completely custom file directories. This represents an enormous quality-of-life improvement for users running Anki portably off external USB flash drives:

1. It eliminates hardcoded local directory write dependencies.

2. It prevents accidental file generation inside hidden application data folders on foreign guest hosts.

OS-Specific Infrastructure Enhancements

• Windows: Patches critical applink startup parameters and display scaling anomalies that plagued the console overlay on older Windows 10 builds.

• macOS: Rewrites launcher permissions to eliminate frustrating developer tool prompts and architecture errors triggered by corrupted Xcode installation hooks on Apple Silicon.

• Linux: Introduces an experimental mode enabling the application window to drop secondary dependencies and hook directly into native system Qt installs, restoring complete compatibility with advanced input methods like Fcitx.

  Linux Magazin  [ News ] 

3. Crucial Security Patches: The 25.09.x Maintenance Cycle

Because Anki renders flashcard templates via HTML, CSS, and embedded JavaScript execution nodes, maintaining absolute security insulation between web engines and host storage is critical. The 25.09.x point releases fixed major zero-day vulnerabilities discovered in the wild.

Media Server Cross-Origin Sanitization

Anki runs a localized media server under the hood to stream images, audio layers, and custom scripts to your cards. Prior to 25.09, this engine did not adequately validate external requests. If a user browsed a malicious webpage via browser clients that lacked strict local isolation (such as older or unhardened variations of Firefox and Safari), the hostile script could theoretically cross-examine local directories while Anki was active. This release strictly hardens request validation policies across the local loopback interface.

Unsafe .apkg Container Restrictions

When importing community-sourced shared decks via .apkg or .colpkg archives, malicious actors could construct directory traversal patterns embedded inside the dataset. If imported on unpatched instances, these packages could compromise the system sandboxing layer and gain read permissions for arbitrary files on the local host machine. Version 25.09 constructs a total cryptographic isolation layer around package extraction and parsing, neutralizing directory path manipulation.

4. Installation & First-Use Operational Workflow

If you are transitioning to the 25.09 platform architecture, the following onboarding map ensures your internal data sets align without fracturing existing database linkages.

Step 1: Clean Installation & Binaries Deployment

1. Head to the official distribution repository or AnkiWeb.

2. Download the unified executable for your target architecture.

3. Upon initialization, a clean terminal interface (the Anki Launcher) will override background steps to establish structural cache parameters, download localized mirror binaries (with new mirror channels added for networks under severe firewall restrictions), and initialize the environment.

Step 2: FSRS Parameter Optimization

1. Once the primary database dashboard populates, open the options wheel next to any top-tier parent deck.

2. Scroll to the FSRS section. If you haven't enabled FSRS yet, toggle it on.

3. Locate your global optimization configuration. Instead of inputting generic global values, look at your primary subdecks. You will notice a newly added, localized fields overlay designed to pass specific overrides directly to individual subsets.

Step 3: Setting Target Thresholds Using "Help Me Decide"

1. Click Help Me Decide located directly beneath the Desired Retention field box.

2. In the modal simulation interface that loads, keep your standard "Days to Simulate" target set to 365 days to measure a sustainable annualized workload.

3. Click Simulate. The program will run thousands of micro-iterations tracking your historical retrieval curves.

4. Review the generated curve. Identify the inflection point where the required study time shifts from a linear angle to an exponential incline. Select a Desired Retention value resting just below that vertical surge to keep your workflow efficient.

Step 4: Synchronizing Cross-Platform Clients

If you study across mobile platforms, ensure your mobile versions are updated to match the 25.09 core logic. For AnkiDroid, update to the latest releases that support the parallel SvelteKit-based UI frontend overhaul. This architecture shares the new JavaScript API layer, ensures proper synchronization across layout components, and natively tracks whiteboard visibility states across active study blocks.

[Read and  Print] [Anki File to Import]