Open Research
Proof, not promises.
We publish our benchmarks, architectural decisions, and performance data. Every claim is backed by reproducible results.
21.2%
Cache Hit Rate
Requests served without touching the LLM
3.3%
Token Savings
Net reduction in input tokens at 100 conversations
<1ms
Cache Latency
Response time for exact cache hits via KV
0
Errors
100% success rate across all benchmark runs
Published Reports
Deep-dives into architecture, benchmarks, and methodology.
2 reports
BenchmarkFebruary 11, 2026by Decyra Research
100-Conversation Benchmark: Decision Replay Performance
Head-to-head comparison of direct OpenAI calls vs Decyra proxy across 100 multi-turn agentic conversations with 8 financial tools.
Read report
3.3%
Token Savings
21.2%
Cache Hit Rate
0
Errors
Research PaperFebruary 8, 2026by Rishabh Singh
Decision Replay Architectures for Agentic AI
Semantic caching, ML-based action routing, and scalable edge inference — a rigorous evaluation of approaches for reducing LLM cost in agentic workflows.
Read report
96%
Feature Coverage
<1ms
L1 Cache Latency
41KB
Bundle Size
See the savings for yourself
Integrate Decyra in 3 lines of code. Watch your AI agent costs drop as the cache warms — backed by the same methodology in our published benchmarks.