5 ms·

Everybody's out here chasing SOTA, meanwhile I'm getting all my coding done with MiniMax M2.5 in multiple parallel sessions for $10/month and never running into

Everybody's out here chasing SOTA, meanwhile I'm getting all my coding done with MiniMax M2.5 in multiple parallel sessions for $10/month and never running into limits.

Aurornis 6mo ago
For serious work, the difference between spending $10/month and $100/month is not even worth considering for most professional developers. There are exceptions like students and people in very low income countries, but I’m always confused by developers with in careers where six figure salaries are normal who are going cheap on tools. I find even the SOTA models to be far away from trustworthy for anything beyond throwaway tasks. Supervising a less-than-SOTA model to save $10 to $100 per month is not attractive to me in the least. I have been experimenting with self hosted models for smaller throwaway tasks a lot. It’s fun, but I’m not going to waste my time with it for the real work.
- zozbot234 6mo ago
  You need to supervise the model anyway, because you want that code to be long-term maintainable and defect free, and AI is nowhere near strong enough to guarantee that anytime soon. Using the latest Opus for literally everything is just a huge waste of effort.
  - dandaka 6mo ago
    Waste of effort... of Opus? If "Opus effort" is cheaper, than dev hours managing yourself more dumb/effective model, what is the point?
    - cyanydeez 6mo ago
      rich people dont concern themselves with the cost of tokens.
      - dnnddidiej 6mo ago
        It is not even rich. If you earn more than $30k it is worth your employer spending $3k on AI tools.
        solenoid0937 6mo ago
        HN is very stingy for some reason, most people here make well over $150k and yet they are still worrying about $200/mo over $10/mo when the former will save more time.
  - senordevnyc 6mo ago
    Yes, but I find supervision much easier and faster with a strong model. It makes fewer dumb mistakes that I have to catch and correct, and it’ll follow my instructions more reliably.
    - jatins 6mo ago
      Depends on the task. If it's something that occurs a lot in training data like React/tailwind code then I don't think you need SOTA. Most reasoning models since Sonnet 3.5, Deepseek 3.1 et al will do fine for those tasks. Doesn't justify 10x the cost in that case imo
- AnonymousPlanet 6mo ago
  For actually serious work, it's a stark difference if your proprietary and security relevant code is sent abroad to a foreign, possibly future hostile country, or is sent to some data center around the corner. It doesn't even need to be defence related.
  - flatline 6mo ago
    AFAIK all these companies have SOTA or near-SOTA models available under enterprise licenses. AI companies are not interested in your secret sauce, they are trying to capture the SDLC wholesale.
    - AnonymousPlanet 6mo ago
      If an American company, let's say a company that writes software for power stations, would use the services of a French or Chinese AI company under such enterprise licenses, how long would you think it would take until someone, in Congress e.g., would interfere? What if they learned that half of the American small and medium sized companies would have started pouring all their business information into such a service?
    - hedora 6mo ago
      I’m not sure what you are implying by “enterprise license”, but if you think it provides any meaningful protection against malicious US government actors, you really need to read and internalize the US CLOUD Act. On a related note, I really need to try some local models (probably starting with qwen), since, at least in 2026, the Chinese models are way better at protecting democracy and free speech than the US models.
    - dnnddidiej 6mo ago
      That doesn't address the concern. Google isn't interested in violating 1st and 4th amendment rights of people who criticize the government... but they do anyway (or more correctly assist the government in doing so).
- slopinthebag 6mo ago
  $100 / month will get you rate limited to much to rely on with the Claude plans. People still report getting rate limited on the $200 / plan. Also not everyone wants to use Claude Code, so if they're paying API pricing it's more likely thousands of dollars a month. If you can get the same results by spending a fraction of that, why wouldn't you?
  - gck1 6mo ago
    And people report getting limited on the $200 plan is putting it very mildly. You can't do any serious work on it without rationing your work and kneecapping your workflows, to the point where you design workflows around anthropic usage limit woodoo rather than what actually works. Without this, I run into WEEKLY usage limits on $200 plan, working on a single codebase, one feature at a time, on just day 3.
    - slopinthebag 6mo ago
      Thats crazy to pin your entire workflow on. Sorry boss, I can't work today I'm being rate limited by Anthropic :/
  - esperent 6mo ago
    I got rate limited within an hour on the $200 while working on a single feature. That was the breaking point, I cancelled my subscription. As it happens I had a low coding workload over the past two weeks so I've been noodling around in PI mostly with Gemini Flash api. I like it - I even agree it's a much better harness than CC. However, the lock in is real. Even without switching models which each have their own quirks, I expect my work speed to drop drastically for at least a week or two even if I was focused on it fully. But after the learning period I think pi will be faster. The danger of course is that CC is fairly on rails while with PI you could end up spending all your time tinkering with the harness.
  - chillfox 6mo ago
    Managing context size and efficient token usage is a skill. I have an Anthropic API key for work, and if I use sonnet/opus all day for agent coding, it ends up costing about ~$25. I am going to need more cpu/ram to run multiple agents in parallel to spend much more than that.
- 0xbadcafebee 6mo ago
  You don't magically get better results by spending 10x more on a model. If your prompt is crap and harness is crap, you get crap results, regardless of model. And if you run into limits, you aren't working at all. Buying the most expensive circular saw doesn't get you the best woodworking, but it is the most expensive woodworking.
  - itake 6mo ago
    Not really true. Remember the prompt engineering craze a few years ago with crazy complex prompt composers (langchain) that don’t need to exist any more because the underlying model got so much better at understanding what the humans are actually asking for?
    - 0xbadcafebee 6mo ago
      A model cannot read your mind. It can guess, and those guesses are more likely to be wrong if you don't give it the right input, and model performance gets worse if not steered/curated properly. The output depends on the input. https://medium.com/@adambaitch/the-model-vs-the-harness-which-actually-matters-more-59dd3116bb31 https://medium.com/@adambaitch/the-model-vs-the-harness-whic... | https://aakashgupta.medium.com/2025-was-agents-2026-is-agent-harnesses-heres-why-that-changes-everything-073e9877655e https://aakashgupta.medium.com/2025-was-agents-2026-is-agent... | https://x.com/Hxlfed14/status/2028116431876116660 https://x.com/Hxlfed14/status/2028116431876116660 | https://www.langchain.com/blog/the-anatomy-of-an-agent-harness https://www.langchain.com/blog/the-anatomy-of-an-agent-harne... (I don't think anecdotes are useful in these comparisons, but I'll throw mine in anyway: I use GPT-5.4, GPT-5.3-Codex, Gemini-3-Pro, Opus, Sonnet, at work every week. I then switch to GLM-5.1, K2-Thinking. Other than how chatty they get, and how they handle planning, I get the same results. Sometimes they're great, sometimes I spent an hour trying to coax them towards the solution I want. The more time I spend describing the problem and solution and feeding them data, the better the results, regardless of model. The biggest problem I run into lately is every website in the world is blocking WebFetch so I have to manually download docs, which sucks. And for 90% of my coding and system work, I see no difference between M2.5 and SOTA models, because there's only so much better you can get at writing a simple script or function or navigating a shell. This is why Anthropic themselves have always told people to use Sonnet to orchestrate complex work, and Haiku for subagents. But of course they want you to pay for Opus, because they want your money.)
- columk 6mo ago
  But $10/mo this year gets you the same quality and throughput as $200/m did last year. Most of us prefer to spend that $190 on our friends and family each month. It's like buying a new iPhone every year - not necessary unless you absolutely need the latest features. If your employer wants you to use a SOTA model they can pay for it.
fnetisma 6mo ago
[dead]
chatmasta 6mo ago
Who are you paying $10/month? OpenRouter?
- tgrowazay 6mo ago
  https://platform.minimax.io/docs/guides/pricing-token-plan https://platform.minimax.io/docs/guides/pricing-token-plan
- 0xbadcafebee 6mo ago
  OpenCode Go, BlackBox, Chutes. https://codeberg.org/mutablecc/calculate-ai-cost/src/branch/main/subscription_vs_api_comparison.csv https://codeberg.org/mutablecc/calculate-ai-cost/src/branch/...
  - chatmasta 6mo ago
    I find Chutes very intriguing… has anyone used it? I found it when I started wondering what sort of $/performance I could get by simply renting GPU machines by the hour and running my own inference.
xutopia 6mo ago
How do you use this? Do you use opencode or another frontend?
- 0xbadcafebee 6mo ago
  yep, OpenCode with a few plugins (context management, memory, a few MCPs)
  - ramon156 6mo ago
    Could you provide a bit more detail? maybe a blog post even? :) I already use opencode and GLM 5.1, I just never really did any research regarding memory, context management, MCP and how to do this efficiently. Would love to hear from people that have got a good setup.

Everybody's out here chasing SOTA, meanwhile I'm getting all my coding done with MiniMax M2.5 in multiple parallel sessions for $10/month and never running into limits.

Aurornis 6mo ago

For serious work, the difference between spending $10/month and $100/month is not even worth considering for most professional developers. There are exceptions like students and people in very low income countries, but I’m always confused by developers with in careers where six figure salaries are normal who are going cheap on tools. I find even the SOTA models to be far away from trustworthy for anything beyond throwaway tasks. Supervising a less-than-SOTA model to save $10 to $100 per month is not attractive to me in the least. I have been experimenting with self hosted models for smaller throwaway tasks a lot. It’s fun, but I’m not going to waste my time with it for the real work.

zozbot234 6mo ago
You need to supervise the model anyway, because you want that code to be long-term maintainable and defect free, and AI is nowhere near strong enough to guarantee that anytime soon. Using the latest Opus for literally everything is just a huge waste of effort.
- dandaka 6mo ago
  Waste of effort... of Opus? If "Opus effort" is cheaper, than dev hours managing yourself more dumb/effective model, what is the point?
  - cyanydeez 6mo ago
    rich people dont concern themselves with the cost of tokens.
    - dnnddidiej 6mo ago
      It is not even rich. If you earn more than $30k it is worth your employer spending $3k on AI tools.
      - solenoid0937 6mo ago
        HN is very stingy for some reason, most people here make well over $150k and yet they are still worrying about $200/mo over $10/mo when the former will save more time.
- senordevnyc 6mo ago
  Yes, but I find supervision much easier and faster with a strong model. It makes fewer dumb mistakes that I have to catch and correct, and it’ll follow my instructions more reliably.
  - jatins 6mo ago
    Depends on the task. If it's something that occurs a lot in training data like React/tailwind code then I don't think you need SOTA. Most reasoning models since Sonnet 3.5, Deepseek 3.1 et al will do fine for those tasks. Doesn't justify 10x the cost in that case imo
AnonymousPlanet 6mo ago
For actually serious work, it's a stark difference if your proprietary and security relevant code is sent abroad to a foreign, possibly future hostile country, or is sent to some data center around the corner. It doesn't even need to be defence related.
- flatline 6mo ago
  AFAIK all these companies have SOTA or near-SOTA models available under enterprise licenses. AI companies are not interested in your secret sauce, they are trying to capture the SDLC wholesale.
  - AnonymousPlanet 6mo ago
    If an American company, let's say a company that writes software for power stations, would use the services of a French or Chinese AI company under such enterprise licenses, how long would you think it would take until someone, in Congress e.g., would interfere? What if they learned that half of the American small and medium sized companies would have started pouring all their business information into such a service?
  - hedora 6mo ago
    I’m not sure what you are implying by “enterprise license”, but if you think it provides any meaningful protection against malicious US government actors, you really need to read and internalize the US CLOUD Act. On a related note, I really need to try some local models (probably starting with qwen), since, at least in 2026, the Chinese models are way better at protecting democracy and free speech than the US models.
  - dnnddidiej 6mo ago
    That doesn't address the concern. Google isn't interested in violating 1st and 4th amendment rights of people who criticize the government... but they do anyway (or more correctly assist the government in doing so).
slopinthebag 6mo ago
$100 / month will get you rate limited to much to rely on with the Claude plans. People still report getting rate limited on the $200 / plan. Also not everyone wants to use Claude Code, so if they're paying API pricing it's more likely thousands of dollars a month. If you can get the same results by spending a fraction of that, why wouldn't you?
- gck1 6mo ago
  And people report getting limited on the $200 plan is putting it very mildly. You can't do any serious work on it without rationing your work and kneecapping your workflows, to the point where you design workflows around anthropic usage limit woodoo rather than what actually works. Without this, I run into WEEKLY usage limits on $200 plan, working on a single codebase, one feature at a time, on just day 3.
  - slopinthebag 6mo ago
    Thats crazy to pin your entire workflow on. Sorry boss, I can't work today I'm being rate limited by Anthropic :/
- esperent 6mo ago
  I got rate limited within an hour on the $200 while working on a single feature. That was the breaking point, I cancelled my subscription. As it happens I had a low coding workload over the past two weeks so I've been noodling around in PI mostly with Gemini Flash api. I like it - I even agree it's a much better harness than CC. However, the lock in is real. Even without switching models which each have their own quirks, I expect my work speed to drop drastically for at least a week or two even if I was focused on it fully. But after the learning period I think pi will be faster. The danger of course is that CC is fairly on rails while with PI you could end up spending all your time tinkering with the harness.
- chillfox 6mo ago
  Managing context size and efficient token usage is a skill. I have an Anthropic API key for work, and if I use sonnet/opus all day for agent coding, it ends up costing about ~$25. I am going to need more cpu/ram to run multiple agents in parallel to spend much more than that.
0xbadcafebee 6mo ago
You don't magically get better results by spending 10x more on a model. If your prompt is crap and harness is crap, you get crap results, regardless of model. And if you run into limits, you aren't working at all. Buying the most expensive circular saw doesn't get you the best woodworking, but it is the most expensive woodworking.
- itake 6mo ago
  Not really true. Remember the prompt engineering craze a few years ago with crazy complex prompt composers (langchain) that don’t need to exist any more because the underlying model got so much better at understanding what the humans are actually asking for?
  - 0xbadcafebee 6mo ago
    A model cannot read your mind. It can guess, and those guesses are more likely to be wrong if you don't give it the right input, and model performance gets worse if not steered/curated properly. The output depends on the input. https://medium.com/@adambaitch/the-model-vs-the-harness-which-actually-matters-more-59dd3116bb31 https://medium.com/@adambaitch/the-model-vs-the-harness-whic... | https://aakashgupta.medium.com/2025-was-agents-2026-is-agent-harnesses-heres-why-that-changes-everything-073e9877655e https://aakashgupta.medium.com/2025-was-agents-2026-is-agent... | https://x.com/Hxlfed14/status/2028116431876116660 https://x.com/Hxlfed14/status/2028116431876116660 | https://www.langchain.com/blog/the-anatomy-of-an-agent-harness https://www.langchain.com/blog/the-anatomy-of-an-agent-harne... (I don't think anecdotes are useful in these comparisons, but I'll throw mine in anyway: I use GPT-5.4, GPT-5.3-Codex, Gemini-3-Pro, Opus, Sonnet, at work every week. I then switch to GLM-5.1, K2-Thinking. Other than how chatty they get, and how they handle planning, I get the same results. Sometimes they're great, sometimes I spent an hour trying to coax them towards the solution I want. The more time I spend describing the problem and solution and feeding them data, the better the results, regardless of model. The biggest problem I run into lately is every website in the world is blocking WebFetch so I have to manually download docs, which sucks. And for 90% of my coding and system work, I see no difference between M2.5 and SOTA models, because there's only so much better you can get at writing a simple script or function or navigating a shell. This is why Anthropic themselves have always told people to use Sonnet to orchestrate complex work, and Haiku for subagents. But of course they want you to pay for Opus, because they want your money.)
columk 6mo ago
But $10/mo this year gets you the same quality and throughput as $200/m did last year. Most of us prefer to spend that $190 on our friends and family each month. It's like buying a new iPhone every year - not necessary unless you absolutely need the latest features. If your employer wants you to use a SOTA model they can pay for it.

fnetisma 6mo ago

[dead]

chatmasta 6mo ago

Who are you paying $10/month? OpenRouter?

tgrowazay 6mo ago
https://platform.minimax.io/docs/guides/pricing-token-plan https://platform.minimax.io/docs/guides/pricing-token-plan
0xbadcafebee 6mo ago
OpenCode Go, BlackBox, Chutes. https://codeberg.org/mutablecc/calculate-ai-cost/src/branch/main/subscription_vs_api_comparison.csv https://codeberg.org/mutablecc/calculate-ai-cost/src/branch/...
- chatmasta 6mo ago
  I find Chutes very intriguing… has anyone used it? I found it when I started wondering what sort of $/performance I could get by simply renting GPU machines by the hour and running my own inference.

xutopia 6mo ago

How do you use this? Do you use opencode or another frontend?

0xbadcafebee 6mo ago
yep, OpenCode with a few plugins (context management, memory, a few MCPs)
- ramon156 6mo ago
  Could you provide a bit more detail? maybe a blog post even? :) I already use opencode and GLM 5.1, I just never really did any research regarding memory, context management, MCP and how to do this efficiently. Would love to hear from people that have got a good setup.