Ask HN: Did Google kill its enterprise workhorse model? | Hacker NewsHacker Newsnew | past | comments | ask | show | jobs | submitloginAsk HN: Did Google kill its enterprise workhorse model?8 points by waldrews 1 hour ago | hide | past | favorite | 1 commentIs anyone else in a panic over the Gemini 2.5 model generation (Pro, Flash) being sunset in October before there's even any Pro class model in general availability (with geo restrictions etc.)? Google wants everyone to migrate to 3.x Flash, which beats the older Pro models on the benchmarked tasks, but isn't the same thing as the Pro class on reasoning-heavy tasks like complex reasoning on very large documents (my big use case).The Gemini family had a distinct niche in document comprehension, with thousand page input documents taking only 300k tokens. Nothing quite like that in OpenAI or Anthropic world, even at more than 10x the token adjusted price. Should we just give up on Google at this point and engineer around the competitors' limits and eat the costs? Totally unnecessary own goal by team Google. help

kennywinker 4 minutes ago [–]
Sounds like they did. IMO, building your business on anything but open-weight models is a bad idea. Unless you're on the s&p 500 you are an insect to google, anthropic, grok (ew), and openai - and they could crush you at any time without even noticing.reply

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact
Search:

There is an expressed concern among users regarding the discontinuation of the Gemini 2.5 model generation, specifically the Pro and Flash versions, slated for sunsetting in October, which precedes the general availability of new Pro-class models with geo-restrictions. This situation raises questions about Google's strategic direction and its impact on specialized model capabilities. The core tension lies in the perceived shift toward migrating users to the 3.x Flash models, which demonstrate superior performance on standard benchmarks, yet do not replicate the capabilities of the older Pro models, particularly concerning complex reasoning over very large documents—a critical use case for some users.

Previously, the Gemini family held a distinct advantage in document comprehension, demonstrated by its ability to process thousand-page input documents using only 300k tokens. This specialized function offered a unique niche compared to offerings from competitors such as OpenAI or Anthropic, even considering the token-adjusted pricing structures of those entities. The discussion revolves around the dilemma of whether users should accept this transition and attempt to compensate by engineering solutions around the limitations of competing models and their associated costs, or if this represents an unnecessary maneuver by Google.

A related perspective suggests a broader caution: building a business foundation on proprietary, closed-weight models is inherently risky. One commentator noted that unless one is operating within a specific tier of market capitalization, reliance on these models subjects entities to potential control by major technology players, including Google, Anthropic, Grok, and OpenAI, potentially leading to vulnerability. This implies that the user community is grappling not only with model performance but also with the broader economic and strategic power dynamics inherent in relying on these foundational models.