One thing about these numbers that's absolutely shocking to me is how low the energy use is:
> That model’s usage was well within our budget ($68, about 4kWh of energy use / 365 grams of carbon emissions).
The energy cost is literally 1% of the total cost. For context, 4kWh of energy would drive you about 15 miles in an EV, about half of the average person's driving miles. It's boiling 10 gallons of water.
With the talk of AI Data Center's impact on the world, you'd think this would be 10x to 100x the amount of energy in order to get the effects they're using here.
My takeaway: the AI data center buildout is an overbuild probably at least as large as the fiber buildout that left us with so much dark fiber. If not even bigger. The only thing that will save the economy is the inability of NVIDIA and chip fabs to produce enough chips to match the buildout planned.
ThibWeb 17 minutes ago [-]
I agree but it does worry me how fast my usage is increasing. Two months ago I was using 10x less tokens and probably not much more than 5kWh on inference. This month about 30kWh on inference. If it becomes more affordable, is there going to be another jump? Not quite sure
monksy 13 minutes ago [-]
GLM5.3-flash has been fantastic for me to make minor fixes in ambigious ways. "Fix x feature, whats going wrong. " It does the job.
surgical_fire 8 minutes ago [-]
GLM-5.3-flash is my implementation model after GLM-5.3 writes the plan.
It's an excellent workhorse. When I am running out of my GLM quota I switch GLM-5.3-flash to DS-4.1-flash.
aktenlage 47 minutes ago [-]
> Unfortunately there are still consequences to it. I chose the 'wrong' model for the prototype, and we spent 450M tokens / $150 / 5kWh of energy use almost overnight. The MCP server itself works well and we now have a great demo of the capabilities, so it’s not for nothing:
> Nonetheless, it’s a good reminder to be careful with model selection and with agentic patterns. We could have achieved similar results for most likely 5x less cost with not that much more effort. Lessons learned! We need to budget for this, and be more careful. Could have seen it coming, but now we know.
I don't get it. Why was it wrong? Which one would have been better? What was the lesson and how could you have foreseen it?
ThibWeb 36 minutes ago [-]
Hmm I might need to rephrase. Initial challenge was to use GLM 5.3 Flash and I was on the non-Flash version for the whole vibe coded build. Just wasn’t paying attention and I didn’t realize that one session was a quarter of the month’s spend and 150% of the budget (big price difference between models)
disiplus 16 minutes ago [-]
Was this post generated with LLM, did he properly mention anywhere why exactly did it fail with example or i have trouble reading.
gpugreg 42 minutes ago [-]
How did you measure energy usage?
Edit: I found a linked article that mentions the inference provider who does the measurements.
ThibWeb 34 minutes ago [-]
Yes, all from Neuralwatt, GPU energy use only. Makes models’ “efficiency” much more visible than tokens.
ThibWeb 5 hours ago [-]
It was a bit of a silly challenge, wasn’t sure how workable, learned a lot in the process about what actually drives usage / costs, and how to keep both under control
cvburgess 4 hours ago [-]
Considering they were your top two models, how did the flash and non-flash versions compare? Did you use them for different tasks?
ThibWeb 31 minutes ago [-]
I have a hard time justifying GLM 5.3 these days. It’s slightly better than Flash but rarely enough to justify the much steeper price. We chose to use usage-based billing only so are very sensitive to model price.
gunalx 48 minutes ago [-]
When text wuality or for pure but adwansed coding is concerned i always pick glm 5.3. The flash is awesome for everything that dosent really matter though.
sampullman 1 hours ago [-]
Worth a comparison with DeepSeek v4.1 flash, if you've got another month to spare!
nowittyusername 15 minutes ago [-]
Ill tell you from my personal use glm 5.3 flash was better then deepseek 4.1 flash, also deepseek liked to yap in his reasoning traces soo fucking much, the yapping was fast but the task was so slow to complete...
ThibWeb 27 minutes ago [-]
Yep, I think next month will be on that. It feels slightly better from a few days of use, and in our WIP benchmarking it scores way higher
samtheprogram 44 minutes ago [-]
DS v4.1 Flash is roughly equivalent. It's going to get some things right/better that GLM flash doesnt and vice versa.
lnenad 53 minutes ago [-]
They've got an image of their homegrown benchmark in the post that lists DS4.1.
throw930rmdkdk 1 hours ago [-]
Flash is pretty decent coder, but it should be paired with good planner and reviewer. I would pick astra low for planning and sol 6.1 medium for reviews.
eikenberry 1 hours ago [-]
What would you use if you wanted to stay (at least) open weight?
verdverm 52 minutes ago [-]
qwen3.8, kimi3, kimi2.7, GLM-5.3 are all good families I use in my coding team
I'm mainly using flash varients, at least as the default, bump.up to stronger model as needed (less often these days)
Geof25 42 minutes ago [-]
qwen3.8 27B or 2.4T? they are completely different models with completely different pricing
verdverm 37 minutes ago [-]
I use all the qwen!
It's my favorite model family to interact with, it's prose is the best imo, it makes me laugh from time-to-time (like when it said it would "crib" some code from another project, lul)
I currently have qwen-flash working on an NES emulator harness so qwen-little can play my first RPG (ff1)
(tho I have used all the others I mentioned, happenstance I'm using qwen this iteration/task)
esafak 55 minutes ago [-]
Deepseek 4.1 Flash and Mimo 2.6 Flash.
esafak 55 minutes ago [-]
Flash is plenty good for planning and reviewing, for my needs. In fact, I use it for that because it's too slow for execution, despite the name.
edit: I subscribe to z.ai, I don't host.
jminnl 42 minutes ago [-]
> I use it for that because it's too slow for execution
What kind of hardware and what particular quant?
gunalx 47 minutes ago [-]
Will agree on this. From z.ai i have found the non flash to have way more consistent performance.
Rendered at 20:09:01 GMT+0000 (Coordinated Universal Time) with Vercel.
> That model’s usage was well within our budget ($68, about 4kWh of energy use / 365 grams of carbon emissions).
The energy cost is literally 1% of the total cost. For context, 4kWh of energy would drive you about 15 miles in an EV, about half of the average person's driving miles. It's boiling 10 gallons of water.
With the talk of AI Data Center's impact on the world, you'd think this would be 10x to 100x the amount of energy in order to get the effects they're using here.
My takeaway: the AI data center buildout is an overbuild probably at least as large as the fiber buildout that left us with so much dark fiber. If not even bigger. The only thing that will save the economy is the inability of NVIDIA and chip fabs to produce enough chips to match the buildout planned.
It's an excellent workhorse. When I am running out of my GLM quota I switch GLM-5.3-flash to DS-4.1-flash.
> Nonetheless, it’s a good reminder to be careful with model selection and with agentic patterns. We could have achieved similar results for most likely 5x less cost with not that much more effort. Lessons learned! We need to budget for this, and be more careful. Could have seen it coming, but now we know.
I don't get it. Why was it wrong? Which one would have been better? What was the lesson and how could you have foreseen it?
Edit: I found a linked article that mentions the inference provider who does the measurements.
I'm mainly using flash varients, at least as the default, bump.up to stronger model as needed (less often these days)
It's my favorite model family to interact with, it's prose is the best imo, it makes me laugh from time-to-time (like when it said it would "crib" some code from another project, lul)
I currently have qwen-flash working on an NES emulator harness so qwen-little can play my first RPG (ff1)
(tho I have used all the others I mentioned, happenstance I'm using qwen this iteration/task)
edit: I subscribe to z.ai, I don't host.
What kind of hardware and what particular quant?