谷歌的Gemini 3.5 Pro又跳票了，内部还挺内耗。竞争对手都发了新模型，这瓜值得吃。
谷歌 Gemini 3.5 Pro 发布再次推迟，已落后于原定6月的目标。公司为提升编码性能更新训练数据但效果不佳，目前与合作伙伴测试但未公布新日期。内部工程师和研究人员因担忧 Anthropic 的 Claude、OpenAI 的 GPT-5 等竞争者领先而感到沮丧。延迟部分源于谷歌内部多部门（搜索、地图、YouTube）协调需求。竞争对手 Grok 4.5、Meta Muse Spark 1.1 和 Kimi K3 上周均已发布，并在关键编码和智能体基准上超越 Gemini 3.1 Pro。
Google's Gemini 3.5 Pro launch delayed *again*, no…
Google's Gemini 3.5 Pro launch delayed *again*, now months behind the June target.
Bloomberg reports the company has been taking extra time trying to lift its coding performance in particular. This has left engineers, researchers and managers frustrated inside Google, with concerns that rivals like Anthropic and OpenAI are pulling ahead with stronger models.
The bigger drag appears to come from Google's own structure. Multiple layers of stakeholders need to be satisfied because any new model has to slot into search, maps, YouTube and the rest of the product suite. That coordination adds friction that narrower competitors can avoid.
Google updated training data late last month to improve coding skills, but the results were disappointing. They're now testing 3.5 Pro with partners but haven't announced a new release date.
At some point Google needs to just ship it. They'll learn more from real-world use than from another round of internal testing. The delay feels like classic DeepMind-versus-product conflict: the research side wants to release models, the product side demands perfect integration across the entire Google ecosystem.
Meanwhile competitors keep shipping. Grok 4.5 and Meta's Muse Spark 1.1 both launched last week, and Kimi K3 just dropped yesterday. All three are beating Gemini 3.1 Pro on key coding and agentic benchmarks. The market isn't waiting for Google to get it perfect.
https://t.co/H2zCksXGGr