kividbio/kividb is a Redis/Valkey-compatible in-memory database written in Rust. The latest release recorded in the repository is v1.1.2, and the provided material does not separately list what changed in this version. The project description positions it as an alternative to the existing Redis/Valkey, emphasizing fast multithreaded processing and built-in search. The update that can be confirmed here is therefore only the release number; the specific changes cannot be described from the given information alone.
Compatibility with existing clients is provided by supporting the Redis protocols RESP2 and RESP3. According to the material, Redis clients and libraries such as redis-cli can be used as they are. The database adds full-text search plus tag, numeric, and vector search to an in-memory key-value store. Full-text search finds words in documents, while vector search finds similarity between data. For storage it uses a sharded memory design, and for data durability it supports snapshots and an optional append-only file. The author's own measurement reports 21.2 million pipelined GET operations per second in a 32-core environment. That is a producer-reported figure taken from the README, not a performance number guaranteed across all environments.
Interpreting the performance figures requires taking the environment into account. The available evidence is limited to the author's measurements, and it cannot be confirmed whether the same results appear on other hardware. The internal structure of the search features is not described in detail in the material either, so it is hard to judge how deep the implementation goes. Also, since the changelog for v1.1.2 is not given, it is unknown how this release changed search or performance. The advantage that can currently be confirmed is the combination of protocol compatibility and built-in search, and neither the differences between releases nor any performance lead can be concluded from this material alone.