Rigario

@danthaeon @JordanMuirFS @honchodotdev I won't say that I tinkered around that much with their settings. One heavy day I did hit $4. I believe (might be wrong) when self hosting dreaming is the thing that costs the most. But since it seems to target off peak periods a slow full local setup doesn't affect it too much.

@JordanMuirFS @Rigario @honchodotdev yes honcho balance draining quickly. i think hermes lets u set it up so it's not used as frequently

If you've recently installed Hermes Agent and are having memory problems, you need @honchodotdev. It literally makes it go away. Hell, even if you're using openclaw, use honcho.dev. I've tried most of the other popular memory systems, honcho is the customizable, reliable, portable and requires the least tinkering. Use it with all your agents and never regret. I self-host, but their costs are reasonable and its easier to get started.

If you're running through pay per token APIs, I guess that kind of costs would be expected if especially if you're using anthropic stuff. I suggest looking at coding plans from Anthropic or like @MiniMax_AI to keep costs predictable and manageable. I personally use 3 sources of tokens, two different coding plans and open router for metered tokens. My total token bill is about $150 per month.

@danthaeon @Rigario @honchodotdev Would also like to know, I’ve gone through $56 in 8 days just running a single Hermes agent

Honcho specifically? Self hosting is free? Just electricity cost? If run through a cost effective API like maybe grok 4.1? I guess it'll be in the ballpark of a dollar or 2 a day? Using their API would probably be about $2? My estimates. Whole system is fairly efficient, I would say I have moderate coding and heavy research and text use cases.

@Rigario @honchodotdev how do ur costs self hosting compare to their API?

@Da7_Tech @NousResearch For real, Hermes is the only open source harness that improves the ability of the model that you use. The rest results in a reduction.

Tried building my app with OpenClaw for a month — constant crashes, endless issues, and weird execution limits. 2 days with Hermes by @NousResearch on the same model: app done.

Honestly there were quite a few: 1. Silently failing was definitely the big one. Often cron silently fails but the session agent would mock the last bit and deliver a result that was part hallucination part actual work. Really hard to catch. 2. Config changes breaking the whole setup was pretty common. Hermes orchestrator almost fixes this completely, really good at catching and checking work. 3. Not respecting the fallback chain was one that I wasted a few days troubleshooting. i.e. it would say that a provider was not working but that would make no sense because its the provider another agent is using. Honestly the list was pretty long. The nefarious thing about all of that is that is often the band aid involved something in the context which bloated context and burnt tokens. Hermes has improved my token efficiency multifold for sure. I haven't quite quantified it but anecdotally I'm quite far from my usage caps.

@Rigario Relate—first 3 weeks with OpenClaw were all stabilization, not shipping. Built a SessionWatcher Dashboard to see subagent states real-time, caught silent failures that looked like config drift. What was your biggest stability headache?

I've been testing the new hermes 0.6.0 release for the last couple of hours. Hermes full multi-agent works as well as I hoped it would. I've completely moved every agent over to Hermes. I started with openclaw more than 2 months ago. My setup is now all Hermes agent. On just openclaw: I spent 40% of my time on config changes and stability fixes. The rest on real work. On hermes + openclaw: I spent about 20% of my time on config changes and stability fixes. Speed up came because Hermes was really good at fixing openclaw Now fully hermes: I expect to only spend 5% of my time on config changes and only when I implement something new. @NousResearch is really cooking. Do yourself a favor, try it.

Hermes agent is the best open source harness. It has the ability to one shot complex tasks but also all the plumbing needed to actually sustain back and forth conversation. Many platforms handle 1 or the other, rarely both. Its also the only one that feels like an actual upgrade to the provider ran chat interfaces when using the same model. If nothing else, just using a model through hermes agent allows you to own (and build upon) your own data. Don't get left behind, start today.

@jadecap_ @ApexTradeFund Congrats! The best part of it all is the consistency, not 1-2 large trades. Lots of respect. Lots to learn from the jade master.

I just received the BIGGEST single payout in prop firm history from @ApexTradeFund this morning. A lifetime of consistent efforts, battling through the darkest moments of life, self-doubt, depression, anger.. It all happens for a purpose. Keep going. 📈🆙

I estimate Rivian's 3Q 2023F, 4Q23F, 1Q24F P&L losses at -$1.0bn, -$0.9bn and -$0.7bn. Rivian will likely raise more capital in 1H 2024! I have assumed quarterly vehicle growth of 15% and 10% in 4Q23F and 1Q24F (we already know 3Q). I also assume average quarterly vehicle COGS reduction of -15%, -11.5%, and -9.0%. I assume average selling prices to remaind at 2Q23A levels at $88.7k. This would imply modest gross profitability in 1Q24F. Despite the above, Rivian's free cash flow generation will likely stay >$1bn negative (-$1.4bn 3Q23F, -$1.3bn 4Q23F, and -$1.2bn 1Q24F) primarily driven by continued operating losses and likely further working capital (WC) expansion. WC appears to be the lowest hanging fruit to improve Rivian's cash burn rate since capital expenditures appear at a low rate already (lower than Tesla's when Tesla was at similar scale but this can also be a negative sign from long term growth outlook). I have included yesterday's announced $1.5bn convertible debt issuance in 4Q23F. This implies Rivian's cash balance as of quarter end will likely change from $7.9bn (Sep-23F) to $8.1bn (Dec-23F), to $6.9bn by March 2024F. This is predicated on Rivian realising the above mentioned COGS savings! This is paramount and will be the number to watch out for in the upcoming 3Q report. Absent a very prosperous 1H 2024 (econcomic outlook much much better than today) I believe Rivian's next capital raise (debt or equity) will be announced in 1H 2024. Note: Cash balance of <$5bn is not acceptable for a company like Rivian, hence capital is sought above this level as we have seen with yesterday's convertible note issuance (when Rivian still had about $8bn in the bank). My interest expense assumption on the new debt from 4Q23 is highly optimistic (kept in line with previous convertible debt issuance since we don't know the terms of the new note as of writing. But this will not be material)

We're excited to support @BuildOnBase 🔵 Collection.xyz will launch on Base mainnet when it goes live, as a Day 1 launch partner. More to come on this, stay tuned!