Hermes Mixture of Agents (MoA) is one of the most practical AI updates I’ve seen this year — and while the feature only just launched, I’ve been running the panel-of-models pattern behind it for weeks via Fusion and Sakana.
It combines several AI models in parallel into one stronger answer — a panel of experts — so you hit frontier-level quality without waiting for gated models.
Key takeaways
- Hermes Mixture of Agents runs several models in parallel and merges them into one stronger answer.
- It’s a clever way around gated, preview-only models like Fable 5 and GPT-5.6.
- A two-model panel (Opus 4.8 + GPT-5.5) beats either model alone on Hermes Bench — and it’s one command to enable.
What Mixture Of Agents Is
MoA is a virtual model provider inside Hermes. Several reference models each give their view privately, and an aggregator model reads them all and writes the final answer.
It’s a panel of experts versus a single genius: combine several strong models and the panel reliably wins on hard tasks.
Why This Matters Today
The best new models are getting gated — Fable 5 is partner-only and GPT-5.6 is a preview — so frontier access is hard to get.
MoA sidesteps that entirely. Combine the models you have, beat any single one, and skip the waiting list.
Panel Beats Genius: The Numbers
Does it work? On Hermes Bench, an Opus 4.8 aggregator over a GPT-5.5 reference beats either model alone:
- Opus + GPT-5.5 panel (MoA): 0.8202
- Opus 4.8 alone: 0.7607
- GPT-5.5 alone: 0.7412
Combining perspectives genuinely lifts quality on hard tasks — roughly 8% above Opus and 11% above GPT, per Hermes’ own benchmark.
How To Turn It On
Setup is genuinely simple:
- Run
hermes updatefirst - Run
hermes modeland choose the Mixture of Agents provider - Pick a preset (or configure your own in
config.yaml) - Switch anytime with
/model default --provider moaor the/moashortcut
It’s provider-agnostic, so you can plug in any models you like.
Stop Chasing The Model, Build The System
Everyone’s waiting on the next model to change everything. But a mix of today’s models already beats the best single model you can’t even access.
The model is the part you swap; the system is what you own. Build the system instead — that’s the lesson MoA hands you for free.
Where I Run It
I run Mixture of Agents inside my Agent OS, alongside Fusion and Sakana Fugu — three systems on the same panel-of-models idea, all in one dashboard, one click apart.
Everything I build with MoA, Fusion and Sakana sits in one dashboard, a click apart. Want the whole stack done for you, with live coaching where I build model panels with you? It’s inside my AI Profit Boardroom (3,800+ operators). New to Hermes? Start free with my AI Money Lab.
The Best Preset To Start With
Out of the box, the top Hermes Bench performer is an Opus 4.8 aggregator with a GPT-5.5 reference, beating either model on its own.
You can also combine cheaper models and still beat a single expensive one — frontier-level results for a fraction of the cost.
Mixture Of Agents, Fusion And Sakana Fugu
This panel-of-models idea also powers Fusion and Sakana Fugu. They’re three takes on the same principle: combine models to reach near-frontier intelligence.
I keep all three in my Agent OS and switch with a single click, picking the right system for each task.
Who Gets The Most From It
Anyone doing serious work with AI who keeps hitting a single model’s limits will benefit most. MoA pushes past that ceiling without special access or big API bills.
It’s a simple, repeatable way to get better outputs from the models you already have.
Why I Treat This As A System, Not A Tool
Hermes MoA only just launched, but after weeks of running the same panel-of-models pattern via Fusion and Sakana, I see it as part of a system rather than a single feature. A panel of models combined into one answer consistently outperforms any single model on the hard tasks that actually matter.
That’s the whole philosophy behind my Agent OS — own the system, swap the models. MoA, Fusion and Sakana are three expressions of that idea.
What It Costs
More models means more tokens than a single call. But the pay-off is that you can mix cheaper models and still beat one expensive model working alone.
Frontier-level quality without frontier-level access or cost — that’s the trade, and it’s a good one.
Also on my network: this Hermes Mixture of Agents guide on JulianGoldie.com, JulianGoldie.co.uk, GoldStarLinks.
FAQ
What is Hermes Mixture of Agents?
A feature that runs several AI models in parallel and merges their answers into one stronger response.
Why use it over one model?
Frontier models are getting gated; a panel of models you already have can beat any single one with no special access.
Does it really beat a single model?
Yes — on Hermes Bench a panel scored 0.82 vs 0.76 for Opus alone.
How do I enable it?
Run hermes update, then hermes model, and pick the Mixture of Agents provider.
Does it cost more?
More tokens for the extra calls, but you can mix cheaper models and still beat one expensive model alone.
The Bottom Line
Hermes Mixture of Agents gives you frontier-level quality from models you already have, with one command to switch it on.
Stop waiting for the next gated model. Build a panel, own the system, and let it outperform the genius working alone.