Tachyon

Open-source, typo-tolerant full-text search in a single binary.

Tachyon adds production-grade text search — BM25 relevance, typo tolerance, filters, facets, sorting, and autocomplete — to your application in under five minutes. It is not a vector database and not a RAG engine; it does one thing.

docker run -p 8108:8108 ghcr.io/tachyon-search/tachyon:latest

Status: alpha. The API is stable enough to build against and every feature below is tested end to end, but this has not run in production anywhere. See Known limitations before you rely on it — in particular, all data currently lives in memory.

---

Quickstart

Create a collection:

curl -X POST localhost:8108/collections \
  -H 'Content-Type: application/json' \
  -d '{
    "name": "products",
    "fields": [
      {"name": "title",       "type": "text"},
      {"name": "description", "type": "text"},
      {"name": "brand",       "type": "keyword", "facet": true},
      {"name": "price",       "type": "int",     "filter": true, "sort": true}
    ]
  }'

Index documents:

curl -X POST localhost:8108/collections/products/documents \
  -H 'Content-Type: application/json' \
  -d '[
    {"id": "1", "title": "Wireless Mouse", "brand": "Logitech", "price": 2999},
    {"id": "2", "title": "Mechanical Keyboard", "brand": "Razer", "price": 8999}
  ]'

Search:

curl 'localhost:8108/collections/products/search?q=wireless+mouse'

{
  "found": 1,
  "search_time_ms": 0,
  "hits": [
    { "document": { "id": "1", "title": "Wireless Mouse", "brand": "Logitech", "price": 2999 },
      "text_match": 554.788 }
  ]
}

Misspell it and it still works:

curl 'localhost:8108/collections/products/search?q=wirelss+mouse'

---

What it does

Relevance	BM25, per-field boosts, phrase matching, term proximity
Typo tolerance	Damerau-Levenshtein, budget scaled to token length
Filters	=, !=, <, <=, >, >=, ranges, set membership, &&, ||, parentheses
Facets	Counted over the whole result set, not the page
Sorting	Any numeric field plus _text_match, multi-clause
Autocomplete	Prefix + typo tolerant, ordered by popularity
Analytics	Top queries, zero-result queries, latency percentiles
Operations	Prometheus metrics, API key auth, crash-safe writes

Full API reference: docs/api.md. Design and internals: docs/architecture.md.

---

Measured performance

From cargo run --release -p tachyon-bench, on an Apple M-series laptop, over a synthetic catalogue where a one-word query matches 6% of the corpus — far broader than real traffic, and deliberately so, because it is the expensive case.

	100k documents	1M documents	Target
Search p95	3.6 ms	67.6 ms	< 30 ms
Search p99	4.5 ms	68.5 ms	< 60 ms
Autocomplete p95	0.09 ms	0.1 ms	< 5 ms
Indexing	210k docs/sec	161k docs/sec	10k docs/sec
Memory	104 MiB	1.0 GiB	—

Reproduce:

cargo run --release -p tachyon-bench -- --documents 1000000 --queries 1000

Indexing throughput beats the target by more than an order of magnitude. Search meets the latency target at 100k documents and misses it at 1M on this corpus; see Known limitations.

---

Known limitations

Read this before choosing Tachyon.

Everything is held in memory. Writes are durable — they go to a write-ahead log and are replayed on startup — but the memtable is never flushed into an on-disk segment, so RAM use grows with the corpus and startup replays the whole log. At roughly 1.1 KiB per document, 5M documents needs about 5 GiB, above the 2.5 GiB the design targets. The segment writer is the single most important next piece of work: the query engine already reads through a source abstraction that segments plug into, and the commit protocol (state.json, WAL generations, tombstone bitmaps) is built and tested around them.

Broad queries scale linearly. Every matching document is scored. At 1M documents a query matching 6% of the corpus costs ~68 ms at p95. Real catalogues are far more selective — and adding a filter already halves it — but the fix is block-max WAND early termination, which would let the executor skip documents that cannot reach the top-K.

Not yet built: distributed clustering, replication, synonyms, stemming, stop words, highlighting, geo search, and nested documents. All are explicit non-goals for v1.

Analytics are not durable. They are an operational signal and reset on restart.

---

Configuration

Every flag has an environment variable equivalent.

Flag	Environment	Default	Meaning
--listen	TACHYON_LISTEN	0.0.0.0:8108	Listen address
--data-dir	TACHYON_DATA_DIR	./data	Collections, WAL, segments
--sync-interval-ms	TACHYON_SYNC_INTERVAL_MS	0	0 fsyncs every write; higher trades durability for throughput
--max-memtable-docs	TACHYON_MAX_MEMTABLE_DOCS	100000	Flush threshold
--admin-key	TACHYON_ADMIN_KEY	unset	Read/write API key
--search-key	TACHYON_SEARCH_KEY	unset	Read-only API key
--log	TACHYON_LOG	info	tracing filter

With no keys set, every endpoint is open. That is right for local development and wrong for anything reachable from a network; set --admin-key before you expose it.

---

Building from source

Needs Rust 1.85 or newer.

cargo build --release        # binary at target/release/tachyon
cargo test                   # 318 tests
cargo clippy --all-targets

The workspace is layered so each crate depends only on the ones below it:

tachyon-server   REST API, auth, analytics, metrics, the binary
tachyon-engine   collection lifecycle, write path, recovery
tachyon-query    parsing, planning, scoring, ranking
tachyon-index    tokenizer, inverted index, columns, fuzzy matching
tachyon-storage  write-ahead log, on-disk layout, metadata
tachyon-core     schema, values, documents, errors

---

Contributing

Issues and pull requests are welcome. Substantial changes should start as an RFC issue so the design can be discussed before the code is written. See CONTRIBUTING.md.

License

Apache 2.0. See LICENSE.