Seeing a lot of talk about #Hermes vs #OpenClaw.
Study the difference between front-loading overkill context (Hermes), and just-in-time (JIT) context loading (OpenClaw). The reason Hermes "just goes off and does it" and OpenClaw goes back-and-forth asking questions is all down to Hermes front-loading and consuming tons of tokens upfront, often needlessly.
TokenBurnMaxxing versus fine-tuning. Once you fine-tune, they end up kinda the same. ⚖ You can make OpenClaw front-load like Hermes. You can make Hermes JIT like OpenClaw.
There are sandbox/toplogy/infra differences which may push you towards one or the other, but in terms of general agentic behaviour which 99% of peeps are looking for, you need to watch your tokens and not overburn them needlessly IMHO.
Just my two sats. 🤷
#AI #LLM
arfonzo
npub1p2…c60yv
2026-04-24 12:12:26 UTC
Author Public Key
npub1p26skxvgyn6wmxr0fayh7ctf7rvsxy3te2s5hjc3emxnhy39y27qtc60yvPublished at
2026-04-24 12:12:26 UTCEvent JSON
{
"id": "15d925f1a042fa9090d69f6754d573c7895cb020b103d33c9f09850a0ace9f9b",
"pubkey": "0ab50b198824f4ed986f4f497f6169f0d903122bcaa14bcb11cecd3b922522bc",
"created_at": 1777032746,
"kind": 1,
"tags": [
[
"t",
"hermes"
],
[
"t",
"openclaw"
],
[
"t",
"ai"
],
[
"t",
"llm"
]
],
"content": "Seeing a lot of talk about #Hermes vs #OpenClaw.\n\nStudy the difference between front-loading overkill context (Hermes), and just-in-time (JIT) context loading (OpenClaw). The reason Hermes \"just goes off and does it\" and OpenClaw goes back-and-forth asking questions is all down to Hermes front-loading and consuming tons of tokens upfront, often needlessly.\n\nTokenBurnMaxxing versus fine-tuning. Once you fine-tune, they end up kinda the same. ⚖ You can make OpenClaw front-load like Hermes. You can make Hermes JIT like OpenClaw.\n\nThere are sandbox/toplogy/infra differences which may push you towards one or the other, but in terms of general agentic behaviour which 99% of peeps are looking for, you need to watch your tokens and not overburn them needlessly IMHO.\n\nJust my two sats. 🤷\n\n#AI #LLM",
"sig": "a3b19d3c7b88a81bcaf7e09bb4bb633fd356fd3f70ebb9fd55d792b5666677ec5d1a485aa0b7086a68468679d370d33b03825b59729ff7e91027356df43c650c"
}