V3.8 · Optimize

Content-aware compression.
Exact cache reuse.

V3.8 provides an exact cache path, content-aware compression controls, provider-prefix preparation, and release-gated semantic-cache safeguards. Cache reuse and compression results remain workload- and configuration-specific.

Feature 01 · Semantic Cache

Semantic caching is experimental.

Semantic caching is opt-in because a similarity threshold can return the wrong response. The implementation contains threshold-learning machinery, but it is not a production claim until the frozen release proves calibration, invalidation, tenant isolation, and false-hit behavior on a representative corpus.

Measured Hit and false-hit rates per corpus
Isolated Tenant and provider-aware keys
slm cache — semantic mode
# Opt-in experimental path $ slm cache semantic on ↳ Enabled locally; this is not evidence of production precision # Required release gate $ slm cache stats Inspect hits, misses, false-hit evaluation, invalidation, and tenant scope
Feature 02 · Extractive Compression

Code-aware compression that preserves structure.

Safe handling for code and JSON protects structure and may return the original content with no reduction. A lossy compression result can be made reversible by storing its original in the CCR path and using the returned identifier before its TTL expires.

Corpus-specific Reduction may be zero in safe mode
Verified Round-trip invariants per structured path
slm compress
$ slm compress codebase-context.json ↳ Input: measure the supplied corpus ↳ Extracted: function signatures, class defs, imports ↳ Output: content- and mode-dependent ↳ Reversible: run the byte-invariant test for this path
Feature 03 · KV-Cache Prefix Alignment

Prepare stable prefixes, then verify provider reuse.

The prefix-ordering path can improve reuse when the configured provider supports prompt caching and its cache conditions are met. Pricing, minimums, TTLs, and discounts belong to the provider's current contract; measure actual usage instead of publishing a fixed saving.

Provider-measured Prompt-cache eligibility and hit rate
Separated Compression and provider-cache evidence
slm prefix-align
$ slm cache --status ↳ Inspect whether proxy caching is enabled ↳ Verify provider prompt-cache eligibility separately ↳ Compare provider accounting before and after ✓ No fixed hit-rate or cost multiplier without release-linked evidence

Ready to measure repeat work?

Upgrade, enable only the paths you need, and compare cache correctness, token use, latency, and provider accounting on your own workload.

Install the published package See Scale Evidence →