Thread

Qwen3.8 Max now ranked as the best overall model by agentic index

521 points · 327 comments · apitman

eli · 1786043776 · focus · HN ↗
I believe it. It's extremely good at troubleshooting. I gave Qwen and Kimi K3 the same annoying, complicated, intermittent bug to track down. Kimi did a bit better in understanding the existing code, but Qwen built some diagnostic tools and did an excellent statistical analysis on the log data. Qwen got way closer to the truth.
I'm very much looking forward to their forthcoming smaller model Qwen 3.8 releases. A version that can easily run locally would be great.
1. comboy · 1786044448 · focus · HN ↗
  How CLI are you guys using for qwen and kimi?
  1. eli · 1786044745 · focus · HN ↗
    I use <a href="https://pi.dev/" rel="nofollow">https://pi.dev/ which works fine out of the box but is fairly minimal and intended to be customized. There are many extensions.
    OpenCode or oh-my-pi might make more sense if you just want a batteries-included agent. You can also make Claude Code work with other models without too much work, but I think that's asking for headaches.
    trey-jones · 1786045258 · focus · HN ↗
    I used claude with GLM and it's easy to set up, just hard to find the documentation. No headaches really, unless you want to use it against multiple different APIs.
    Gooblebrai · 1786047151 · focus · HN ↗
    Is there any subscription of any kind for Qwen? Or via Pi.dev needs to be used with API credits?
    iAMkenough · 1786047534 · focus · HN ↗
    My first web search turned up this as the top result <a href="https://www.alibabacloud.com/help/en/model-studio/coding-plan" rel="nofollow">https://www.alibabacloud.com/help/en/model-studio/coding-pla...
    adamgoodapp · 1786086875 · focus · HN ↗
    Currently only these models are available: qwen3.7-plus (vision), qwen3.6-plus (vision), kimi-k2.5 (vision), glm-5, and MiniMax-M2.5
    lkt · 1786053136 · focus · HN ↗
    Opencode Go has Qwen 3.8 Max at $10/month
    Gooblebrai · 1786055040 · focus · HN ↗
    Found the usage limits on OpenCode Go quite poor tbh
  2. g58892881 · 1786046149 · focus · HN ↗
    pi
  3. bitexploder · 1786054364 · focus · HN ↗
    OpenRouter and Pi here now.
  4. MrDrMcCoy · 1786075089 · focus · HN ↗
    Hermes. It can be set up to delegate to OpenCode for better multitasking and context usage as well.
2. thefourthchime · 1786045525 · focus · HN ↗
  Did you also try Opus 5 and 5.6 Sol?
  1. ghosty141 · 1786055032 · focus · HN ↗
    5.6 sol was very impressive for me. I had a weird behavior while using Qt and I gave it a screenshot and my expectation of what should happen and it read the Qt sourcode and showed me that my issue was a bug (including link to the ticket).
  2. sscaryterry · 1786056131 · focus · HN ↗
    Opus 5 is just terrible

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.