‹ BackHN Continuity

Thread

Qwen 3.8 Omni Flash

346 points · 138 comments · jjcm

  1. _ache_ · · focus · HN ↗
    If the performances are comparable, and there is no evidence it's not.

    in/out ($) Gemini : 1.5 / 9.0 | Qwen 3.8: 0.15 / 0.47

    That is a massive cost reduction.

    Refs: <a href="https:&#x2F;&#x2F;www.alibabacloud.com&#x2F;help&#x2F;en&#x2F;model-studio&#x2F;model-pricing#china-beijing-h4" rel="nofollow">https:&#x2F;&#x2F;www.alibabacloud.com&#x2F;help&#x2F;en&#x2F;model-studio&#x2F;model-pric... <a href="https:&#x2F;&#x2F;runware.ai&#x2F;gemini-omni" rel="nofollow">https:&#x2F;&#x2F;runware.ai&#x2F;gemini-omni

    1. killingtime74 · · focus · HN ↗
      You can&#x27;t just look at the per token cost, but how many tokens it takes on average to do a task. The difference can be massive.
      1. vntok · · focus · HN ↗
        True, but it would have to be more than massive (order(s) of magnitude) to offset that gap.
        1. kaliqt · · focus · HN ↗
          We notice with frontier models like Astra and Fable that one might use a lot less tokens than the other to complete the task thereby being the better deal in spite of the far higher token cost.
          1. LeBit · · focus · HN ↗
            What is Astra $&#x2F;task?

            Even if OpenAI end up using 1 token for per task, if the token costs 1M$ , some people will find it expensive.

            1. nater5000 · · focus · HN ↗
              <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;#intelligence-comparison-tabs" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;#intelligence-comparison-tabs

              Astra on xhigh has a cost per task of $2.31 with an intelligence index of 53. Qwen3.8 Max has a cost per task of $5.41 with an intelligence index of 45. Pricing for GPT-6 Astra (xhigh) is $10.00 per 1M input tokens and $50.00 per 1M output tokens. Pricing for Qwen3.8 Max (0902) is $2.00 per 1M input tokens and $6.00 per 1M output tokens.

              Obviously this is just one measure of all of this (and Qwen 3.8 Omni Flash isn&#x27;t yet available), but I think this illustrates the point well. These relative task costs are pretty consistent across different analysts. Cost per token is arguably a useless measure at this point in most circumstances.

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.