Ensemble AI

The smartest AI is an ensemble.

Features

Query 12+ AI models simultaneously, analyse their agreement, and synthesise the best answer

What Makes Ensemble AI Different

Agreement Analysis

See exactly where models agree or diverge. Vector embeddings and cosine similarity create an N×N comparison matrix with colour-coded results.

N×N Matrix
Cosine Similarity
Real-time Stats

Dual Consensus Strategies

Standard: synthesise all responses. Elo Ranking: tournament-style LLM judging ranks responses, then synthesises only the top performers.

Elo Rating
Top N Selection
Impartial Judge

Human + AI Integration

Add responses from external tools, human experts, or other AI services. Included in agreement analysis and consensus generation.

Full Integration
Agreement Scoring
Consensus Synthesis

Built for Performance

12+

AI Models

Across OpenAI, Anthropic, Google, and xAI

<100ms

Streaming

Buffered chunks with parallel responses

AES-256

Encryption

GCM mode with random 96-bit IVs

Zero

Lock-in

Direct browser-to-API in Free Mode

From Question to Consensus in 4 Steps

1

Configure

Choose Free Mode with AES-256 encrypted API keys or Pro Mode with managed credits.

2

Select Ensemble

Pick 2–6 models across 4 providers. Choose your summariser and consensus method.

3

Enter Prompt

Write your question. Optionally add manual responses from experts or external tools.

4

Review & Analyse

Watch responses stream in parallel. See agreement analysis and synthesised consensus in real-time.

Two Ways to Use Ensemble AI

Free Mode

BYOK

Bring your own API keys. Direct browser-to-API communication with zero server dependency.

  • AES-256-GCM browser encryption

  • Direct provider API calls

  • Pay only provider token costs

  • Full feature access

  • No usage fees

Pro Mode

Coming Soon

Managed backend with credit-based billing. No API key management needed.

  • Managed API connections

  • Credit-based billing

  • No key management

  • Usage analytics

  • Priority support

12+ Models Across 4 Providers

Access the best models from every leading AI provider. New models appear automatically via dynamic discovery.

OpenAI

GPT-4o, GPT-4o Mini, o1-preview, o1-mini

Anthropic

Claude 3.5 Sonnet, Claude 3 Opus, Claude 3 Haiku

Google

Gemini 1.5 Pro, Gemini 1.5 Flash, Gemini 2.5 Flash

xAI

Grok 2, Grok 2 Mini

Security & Privacy

Military-grade encryption with zero server-side storage in Free Mode.

  • AES-256-GCM encryption via Web Crypto API

  • Device-specific key derivation (SHA-256)

  • Random 96-bit IV per encryption

  • Plaintext keys never touch localStorage

  • No server-side key storage in Free Mode

  • No prompts or responses logged or stored

  • Direct browser-to-provider communication

  • Open-source for full transparency

About Ensemble AI