A drop-in Redis replacement built in Rust. Multi-threaded. AI-Ready.
Faster. Safer. Free.
Redis uses a single CPU core. Your 64-core server? Wasted potential.
Redis Enterprise charges per node. Scale = Massive bills.
Vector Search? Time-Series? Graph? You need expensive add-ons.
Everything Redis has, plus everything it doesn't.
Shared-nothing design. Each CPU core runs its own event loop. Linear scaling with hardware.
Native semantic search with VADD.TEXT auto-embedding. Build RAG apps instantly.
High-frequency metrics with TS.ADD and TS.RANGE. Perfect for IoT & Finance.
Adjacency lists with BFS/DFS traversal. Social networks, fraud detection, recommendations.
Store and query JSON documents natively with JSON.SET and JSON.GET.
O(1) probabilistic membership checks. Process billions of records efficiently.
Zedis speaks HTTP on port 9200. Use your standard ElasticSearch clients without code changes.
Auto-sync data from MySQL, Postgres, SQLite. Configure via zflow.toml. Hot-reload enabled.
Feature-by-feature, benchmark-by-benchmark. Zedis wins.
| Capability | Zedis | Redis |
|---|---|---|
| Price | FREE Forever | $$$ |
| Threading Model | Multi-Threaded (Thread-per-Core) | Single-Threaded |
| Throughput (ops/sec) | 1,000,000+ | 100,000 |
| P99 Latency | <100ΞΌs | ~500ΞΌs |
| Native Vector Search | β Built-in | β Requires RediSearch |
| Time-Series | β Built-in | β Requires RedisTimeSeries |
| Graph Queries | β Built-in | β Requires RedisGraph (deprecated) |
| JSON Documents | β Built-in | β Requires RedisJSON |
| Memory Safety | β Rust (Guaranteed) | β οΈ C (Manual) |
| Auto-Embedding for AI | β VADD.TEXT | β Client-side only |
| ML Inference | β ML.RUN | β Not supported |
Zedis isn't just a cache. It's a multi-model data platform.
Semantic search with cosine similarity. Auto-embed text with VADD.TEXT. Perfect for LLM/RAG pipelines.
High-frequency data ingestion. Range queries with TS.RANGE. Foundation for Gorilla compression.
BFS/DFS traversal with GRAPH.BFS. Model relationships, detect fraud, power recommendations.
Load models with ML.LOAD, run predictions with ML.RUN. Zero data movement.
Drop-in compatible. Use your existing Redis clients.
import redis
# Connect to Zedis (same as Redis!)
r = redis.Redis(host='localhost', port=6379)
# AI: Auto-Embedding Vector Search
r.execute_command('VADD.TEXT', 'products', 'Wireless Noise Cancelling Headphones')
r.execute_command('VADD.TEXT', 'products', 'Bluetooth Speaker Portable')
results = r.execute_command('VSEARCH.TEXT', 'products', 'audio devices', '5')
# Time-Series
r.execute_command('TS.ADD', 'sensor:temp', '1704326400', '25.5')
data = r.execute_command('TS.RANGE', 'sensor:temp', '0', '+')
# Graph
r.execute_command('GRAPH.ADD', 'social', 'alice', 'bob')
friends = r.execute_command('GRAPH.BFS', 'social', 'alice', '2')Replace Redis today. Zero migration cost. Zero license fees.
Available for Windows, Linux, and macOS. MIT Licensed.