TPEE Toptronic® Prompt Engineering Editor

71. Mixture of Agents

Learn how Mixture of Agents uses multiple LLM agents in layers to improve answer quality.

By Jacques Botte, founder of Toptronic®. Last updated 12 September 2026.

The lesson

Modern AI systems often need more than one pass of reasoning. A single model may miss an edge case, make an arithmetic error, or choose the wrong interpretation of an ambiguous request. Mixture of Agents, or MoA, is an architecture that improves answer quality by using multiple LLM agents working together in layers.

MoA is not the same as Mixture of Experts, or MoE. MoE happens inside a single neural network: different internal "expert" sub-networks handle different tokens. MoA happens outside the model, across independent agents that each produce a complete response to the same prompt.

The standard MoA pattern has two layers. In the first layer, several "proposer" agents each generate their own answer to the prompt. Because each proposer may use a different model, temperature, or system prompt, the group produces a diverse set of candidate responses. In the second layer, one or more "aggregator" agents receive all the proposer answers and synthesize the best final response.

The aggregator is given explicit instructions, such as: "Review the following candidate answers and produce a final answer that is more accurate, complete, and concise than any individual candidate." This forces the system to self-correct by comparing alternatives, catching errors that a single model might miss.

MoA is especially useful for complex reasoning, coding, mathematics, safety-critical answers, and tasks where no single model is consistently best. It can also be used to combine a cheap fast model for proposals with a strong model for aggregation, trading cost against quality.

The trade-offs are real. MoA increases latency because multiple models must run. It increases token cost because you pay for every proposer and aggregator call. And it adds operational complexity: you must handle failures, timeouts, and disagreements between agents.

Design tips: keep the number of proposers small (3-5 is common), give each proposer a slightly different angle, make the aggregator's criteria explicit, and always evaluate whether the quality gain is worth the extra cost for your use case.

Check yourself

Question 1: What is the main difference between MoA and MoE?
  1. MoA uses multiple independent agents; MoE uses internal expert layers inside one model — correct
  2. MoA is faster than MoE
  3. MoA only works with OpenAI models
  4. MoE is the same as MoA

Answer: MoA uses multiple independent agents; MoE uses internal expert layers inside one model

Mixture of Agents coordinates multiple independent LLM agents, while Mixture of Experts is an internal neural-network architecture.

Question 2: What do aggregator agents do in an MoA system?
  1. They generate the initial proposals
  2. They combine and improve the proposer outputs into a final answer — correct
  3. They delete bad prompts
  4. They train the models

Answer: They combine and improve the proposer outputs into a final answer

Aggregators receive candidate answers from proposer agents and synthesize a better final response.

Question 3: Which of the following is a valid trade-off when using MoA?
  1. Lower latency
  2. Higher token cost and latency — correct
  3. No need for testing
  4. Guaranteed correct answers

Answer: Higher token cost and latency

Running multiple proposer and aggregator agents increases both token cost and response latency.

← Previous lesson · All 83 lessons · Next lesson →

The full course — 83 lessons and 249 quiz questions — ships inside the app. Get TPEE to study it offline.