Valkey vs Redis in 2026: The Fork Quietly Won Most of the Market

Valkey, a fork of Redis, has quietly become the dominant in-memory data store across most internet platforms since March 2024. This shift occurred due to a licensing change, not superior technology. Redis was originally released under the BSD 3-Clause license, but in March 2024, Redis Ltd. announced a dual licensing model for versions 7.4 and above, which included the Standard Software Patent License (RSALv2) and the Server Side Public License (SSPLv1).

Neither of these licenses is OSI-approved, leading to AWS, Google Cloud, Oracle, Ericsson, and Snap backing the Linux Foundation's fork, Valkey, which maintains a BSD license.

In May 2025, Redis 8.0 reintroduced the AGPLv3 license option, which is OSI-approved, making Redis open source again as of that time. However, the real question in 2026 is whether to choose between Valkey, Redis, or both based on specific workload, legal requirements, and cloud costs.

Valkey has made significant performance improvements since its fork. Valkey 8.1, for instance, redesigned the hash table, reducing 16 bytes of overhead per key, and added pipeline memory prefetching and SIMD acceleration. In March 2026 benchmarks, Valkey 8.1 showed about 8 percent higher single-threaded SET throughput (154,000 ops/s vs Redis 8.2's 142,000 ops/s) and roughly 40 percent higher pipelined SET throughput (812,000 ops/s vs Redis 8.2's 580,000 ops/s).

Valkey also demonstrated a 33 percent reduction in P99 latency under load (2.8 ms vs Redis 8.2's 4.2 ms) and an overall 191 MB memory footprint compared to Redis's 198 MB for 1 million keys with 64B values each.

Despite Valkey's performance edge, the decision to migrate depends on the specific use case. If commands are issued one at a time, the 8 percent difference in throughput is unlikely to alter architecture decisions. However, if batch commands are used in latency-sensitive paths, Valkey's 40 percent pipelined throughput gain can be the deciding factor between scaling up and scaling out.

The most compelling case study is Snap Inc., which reduced its infrastructure costs by 60 percent by migrating from Redis to Valkey. Snap's yearly cost dropped from $2.1 million to $840,000, primarily due to the elimination of licensing fees, fewer node requirements, and improved memory efficiency. While Valkey instances run about 20 percent cheaper than Redis instances on AWS ElastiCache, the cost savings become more significant at higher Redis bills.