Melaya — Build AI agents for any job. Agentic platform for research, ops, outreach, reporting — and the only one where agents can actually trade.

// Benchmarks

Every Melaya number.
One place to verify.

Counts ticked from the source of truth on every build. Latency measured by the engine itself with a monotonic clock. Click any number to see the file, the methodology, and the raw data behind it.

SIGNAL FIELD / LIVE

Six proofs. Every number reproducible.

01 / tools
1,365
public tool functions

Enumerated at build time. Every entry carries an id, category, and a one-line description.

Browse the catalog
02 / subagents
110
persona templates

ReAct agent definitions across 15 crews, grouped by crew with a per-persona role.

Browse the library
03 / ai providers
20
model providers

Bring-your-own keys for hosted providers, plus local-first runners and open routers.

See providers
04 / rust engine
310 ns
p50 · ticker state-cache write

89,033 samples, engine 0.4.48. Full Rust pipeline 14.4 µs end to end.

See the methodology
05 / agentic framework
16
metrics measured

The Python agentic runner, measured end to end: dispatch, RAG, model wrapper, HITL gate, cost, tracing, prompt-injection, crews.

See the methodology
06 / methodology
OSS
every number reproducible

Clone the open-source bench suites and re-measure on your own hardware. No platform access required.

github.com/melaya-labs/melaya
Join the community