Save Thousands on API Costs.
Make Workflows 10x Faster.
SmartCache intelligently caches expensive calls to LLMs, media generation, OCR, and external APIs — across JSON and binary data — so repeated work returns instantly.
The Problem SmartCache Solves
Stop burning money on repeated API calls. SmartCache cuts cost and latency for repeatable work across LLMs, media generation, OCR, and external APIs.
- LLM API Calls: Save $100s–$1000s monthly on GPT/Claude/Gemini runs
- Media Generation: Avoid repeated Imagen/Midjourney/SD charges for identical prompts
- File Processing: Cache expensive PDF parsing, OCR, and audio transcription
- External APIs: Reduce redundant requests and dodge rate limits
- Binary Operations: Reuse processed images, videos, and documents
Key Benefits
Designed for n8n workflows handling JSON, binaries, and high-cost external calls.
- Automatic SHA‑256 cache keys from sorted input
- Universal compatibility: JSON, binaries, API responses
- TTL support (hours or infinite)
- Selective fields for key generation
- Persistent file-based storage
- Dual inputs: Check and Write
- Explicit outputs: Hit and Miss
- Node-specific isolation per instance
- Batch and individual modes
- Clear, structured logging
High‑Value Use Cases
Where SmartCache pays for itself fast.
🧠 LLM & AI Operations
Example: Summarize 1000 articles per run. With ~85–90% hit rate, cost drops from dollars to cents, and results return instantly.
🖼️ Media Generation & Processing
Cache image/video outputs for identical prompts or parameters; avoid minutes-long re-renders and duplicate charges.
📄 Document Processing
OCR, PDF parsing, and transcription are costly and slow — cache once, reuse forever.
🌐 External API Calls
Reduce redundant fetches, manage rate limits, and stabilize response times with cached payloads.
How It Works
Dual input/output design that only processes cache misses, then persists results for future hits.
Flow
- Check: compute key from input, look up cache
- Process: only misses hit APIs/LLMs/expensive steps
- Write: persist results for next-time instant hits
Inputs: Check, Write • Outputs: Hit, Miss
Cache Key Generation
- SHA‑256 hashing for consistent, secure keys
- Node ID included to prevent collisions
- Selective fields for fine-tuned strategies
- Sorted keys to guarantee deterministic hashing
Performance Impact
Gains depend on hit rate and original latency. Typical results:
| Use Case | Typical Hit Rate | Cost Reduction | Execution Time |
|---|---|---|---|
| LLM Article Processing | ~85% | ~6.7× | Seconds → ms |
| Generative AI Media | ~70% | ~33× | Minutes → ms |
| PDF/Document OCR | ~95% | 20× | ~30s/page → <50ms |
| Frequent API Polling | ~90% | 10× | Seconds → ms |
| Audio Transcription | ~80% | 5× | Seconds → ms |
Installation & Configuration
Install via n8n GUI or npm, then configure the node parameters.
Install
n8n (GUI):
- Settings → Community Nodes
- Install →
n8n-nodes-smartcache
Self-hosted:
npm install n8n-nodes-smartcacheNode Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
| Batch Mode | Boolean | false | Process all items as a single unit |
| Force Miss | Boolean | false | Force cache miss and regenerate |
| Cache Key Fields | String | "" | Comma-separated fields for key (empty = all) |
| TTL (Hours) | Number | 24 | 0 = infinite |
| Cache Directory | String | /tmp/n8n-smartcache | Where cache files are stored |
Inputs: Check, Write • Outputs: Hit, Miss
Usage Examples
Importable workflows with previews included in the repo.
Support
Issues, community, and licensing.
- GitHub Issues — report bugs or request features
- Discord — community chat
- License: Mozilla Public License 2.0
Install
Use n8n’s GUI or add the package directly.
npm install n8n-nodes-smartcachen8n GUI: Settings → Community Nodes → Install → n8n-nodes-smartcache