Comparison

Humalike vs Hume.

Hume builds an empathic speech-to-speech model. Humalike sells the behavioral layer above any voice stack — multi-party, theory of mind, persona, memory.

Talk to usVisit Hume

Them

Hume

Empathic voice model

EVI 3 (May 2025): speech-to-speech foundation model with prosody-based turn-taking and any-voice prompting. BYO-LLM (Claude, GPT, Gemini). Plus Octave TTS. Voice-first; Expression Measurement API sunsets June 14, 2026.

Us

Humalike

Behavioral infrastructure

Behavioral primitives — turn-taking, norms, persona, memory, theory of mind, social signals — composed above any voice stack. Voice and text, multi-party, model-agnostic. Hume is a clean voice layer to plug in underneath.

The core differences.

No feature-checklist arms race. Just the dimensions that decide which one fits — with the honest call on each.

Empathic voice model (speech-to-speech)

Hume

EVI 3 — their flagship.

Humalike

Plug Hume in as the voice layer.

Composable behavioral primitives

Hume

EVI is a bundled voice stack.

Humalike

7 APIs you compose.

Multi-human rooms

Hume

EVI is 1:1 over a WebSocket.

Humalike

Multi-party as a first-class case.

Cross-session relational memory

Hume

Chat resumption only — no semantic memory.

Humalike

Memory per person, across sessions.

Text-channel agents

Hume

Voice-first stack.

Humalike

Voice and text.

Theory of mind + social signals

Hume

Prosody only.

Humalike

Beliefs, intent, typing pauses, edits.

Persona grounded in data

Hume

Voice identity, not behavioral persona.

Humalike

Persona from real community/world data.