THIS OPEN-SOURCE AGENT JUST MADE HERMES LOOK SLOW AND EXPENSIVE. SAME TASK, 4X FASTER, A THIRD OF A CENT One prompt in both: research the 10 most popular AI coding agents, compare pricing, features, open-source status and GitHub popularity in a table, include sources. • Speed > OpenHuman (TinyHumans... AI) finished the table in about 25 seconds, 8 tool calls, 10 sources. > Hermes needed over a minute and a half for the same job. • Cost > Total bill: $0.0036, ticking live on screen while it ran. > Context window live too: 37.7K of 128K, broken down by what used it. > Hermes (Nous Research) hands you the answer and hides the meter. • Why it's not a toy > Open source, 39K+ GitHub stars. > Memory lives on your own machine, not in someone's cloud. > Your own API keys or local Ollama models, your call. Speed you can time and cost you can see -> every other agent is asking you to trust it on both. OpenHuman by TinyHumans AI, the Hermes alternative I'm switching to. Download link ↓show more

->

@slash1sol love seeing open-source level up. key thing: make sure the data sources are reliable and up-to-date.

ty

the meter matters when agents run all day

insane tech

yeah that’s true

which agent did you compare? curious if the speedup holds on longer tasks or just that specific research prompt

Compared OpenHuman to Hermes. Speedup was clear on this research task - testing longer ones now.

openhuman basically runs my comparison tasks better than i ever could

🫡🫡

benchmarks or cap?

cheap agents win when the meter stays visible

The perspective shift from this post is welcome

an interesting agent

top tier repo thanks

The speed difference is wild, and we actually went deeper on how training inside real coding harnesses can impact performance:

a live context meter broken down by source is rare

Open source is moving fast. 4x faster at a fraction of the cost is wild

love the live cost meter. but i noticed hermes was checking vendor pricing pages directly while openhuman used comparison blogs, would be cool to see an accuracy check next to the speed one

25 seconds is seriously fast

the live cost meter matters more than the 4x an agent that shows $0.0036 while it runs is one you can actually leave on overnight

The live cost meter is the genuinely useful part. Benchmarks that compare agents on one prompt tend to hide the rerun tax: 8 tool calls looks cheap until a task needs 40, and the numbers stop being comparable across models.

A third of a cent per run is great if the table and sources hold up. That's the part I'd check before trusting the 4x.

4x faster on a research table is easy mode. curious if anyone has tried it on a messy existing codebase yet

Another opinion that only strengthened my with switch to that model

live meter is more useful than it sounds

slash1s cooking

worth using

This is the kind of agent I want: fast, cheap, open source, and transparent. No guessing where the money goes.

if it really delivers that much speed at such a low cost, it's gonna shake things up

Speed you can measure beats speed you have to trust

a third of a cent is kinda absurd. want to see it on a messier task though