Decision Workspace
tokenizer vs text-splitter vs unicode-segmentation
Side-by-side comparison of Rust crates
39
tokenizer
growingv0.1.2
Thai text tokenizer
64
text-splitter
growingv0.33.0
Split text into semantic chunks, up to a desired chunk size. Supports calculating length by characters and tokens, and is callable from Rust and Python.
63
unicode-segmentation
growingv1.13.3
This crate provides Grapheme Cluster, Word and Sentence boundaries according to Unicode Standard Annex #29 rules.
Core Metrics
| tokenizer | text-splitter | unicode-segmentation | |
|---|---|---|---|
| Health Score | 39 | 64 | 63 |
| Total Downloads | 4.1K | 2.4M | 597.6M |
| 30d Downloads | 0 | 0 | 0 |
| Dependents | 0 | 1.0K | 22.3K |
| Releases | 2 | 65 | 27 |
| Last Updated | 2323d ago | 16d ago | 130d ago |
| Age | 6y 4m | 3y 5m | 11y 6m |
Health Breakdown
tokenizer
Maintenance
3
Quality
17
Community
6
Popularity
4
Documentation
9
text-splitter
Maintenance
18
Quality
13
Community
13
Popularity
8
Documentation
12
unicode-segmentation
Maintenance
11
Quality
17
Community
16
Popularity
8
Documentation
11
Technical Details
| tokenizer | text-splitter | unicode-segmentation | |
|---|---|---|---|
| Version | 0.1.2 | 0.33.0 | 1.13.3 |
| Stable (≥1.0) | ✗ No | ✗ No | ✓ Yes |
| License | BSD-3-Clause | MIT | MIT OR Apache-2.0 |
| Dependencies | 2 | 20 | 3 |
| Crate Size | 17KB | 62KB | 112KB |
| Features | 3 | 4 | 1 |
| Yanked % | 0.0% | 1.5% | 7.4% |
| Edition | 2018 | 2021 | 2018 |
| MSRV | — | 1.90.0 | 1.85.0 |
| Owners | 1 | 1 | 5 |
Links
Quick Verdict
- •text-splitter leads with a health score of 64/100, but none of the options score above 80.
- •unicode-segmentation has the most downloads (597.6M), suggesting wider adoption.
- •unicode-segmentation is depended on by 22.3K crates — strongest ecosystem trust.
- •⚠ tokenizer has not been updated in over a year.
- •tokenizer, text-splitter are pre-1.0 — API may change.