Performance Methodology

Transparency is our core value. All performance claims on this site are verifiable, reproducible, and sourced from open benchmarks.

Test Environment

  • Hardware: Apple M4 Max (14-core CPU), 36GB Unified Memory
  • Network: Localhost loopback (127.0.0.1) for baseline latency
  • OS: macOS Sequoia 15.2 (25.2.0)
  • Implementation: talos-core-rs (v0.2.0/a639264d) compiled with --release

View full benchmark suite on GitHub →

Verified Claims

Talos Core authorizes requests at high throughput.

12klow Risk
key: gateway.authorize_fast.metrics.req_per_sec
Source: M4 Max, service-level gateway authorize_fast
Verify Source ↗

Signature verification at microsecond latency.

< 150μslow Risk
key: crypto.verify_ed25519.latency_ms
Source: M4 Max, Ed25519 verification
Verify Source ↗

Double Ratchet encryption at scale.

44k+low Risk
key: crypto.ratchet_encrypt.ops_sec
Source: Python SDK, Session.encrypt(35B)
Verify Source ↗

Canonical JSON digest computation.

270k+low Risk
key: crypto.canonical_json.ops_sec
Source: Python SDK, SHA256 + canonical serialization
Verify Source ↗

Verified large-scale behavior modeling.

12klow Risk
key: audit.log_ingest.metrics.req_per_sec
Source: M4 Max, service-level audit log_ingest
Verify Source ↗

End-to-end authorization latency at scale.

< 1.5mslow Risk
key: gateway.authorize_fast.metrics.p50_ms
Source: M4 Max, service-level p50 latency
Verify Source ↗

Cryptographic proof integrity verified for all authorized actions.

100%low Risk
key: gateway.authorize_fast.metrics.error_rate
Source: Based on 0.0% error rate in gateway authorization benchmarks
Verify Source ↗