@jakevin7: An interesting thing. The DeepSeek V4 technical report conducted a comprehensive evaluation of all major LLMs, concluding that Gemini 3.1 Pro has the strongest world knowledge among all models. Not GPT, not Claude, but Gemini. But when people use Gemini...
Summary
According to the DeepSeek V4 technical report's evaluation of mainstream LLMs, Gemini 3.1 Pro is considered to have the strongest world knowledge, but users generally find it hard to use because the model does not proactively use search tools.
View Cached Full Text
Cached at: 06/08/26, 05:14 AM
There’s an interesting observation.
In its technical report, DeepSeek V4 conducted a comprehensive evaluation of all major mainstream models, concluding that Gemini 3.1 Pro has the strongest world knowledge among all models.
Not GPT, not Claude — it’s Gemini.
Yet the general user experience with Gemini is: does it even work well?
The issue isn’t the model itself; it’s that Gemini is extremely lazy to take action.
If you ask it about the latest news, it has a search tool, but it just won’t use it proactively. Many times you have to explicitly say “go search for it” before it bothers to look. It’s like a well-read person — you ask what’s been happening lately, and they shrug: “I haven’t read today’s newspaper.”
A model with the best world knowledge in existence, yet too lazy to use its tools — that’s the real reason Gemini feels so awkward to use.
Similar Articles
@wquguru: https://x.com/wquguru/status/2057852569054278045
Performed source code analysis and multi-model testing on the pi-goal tool, finding that DeepSeek V4 Pro is 31x cheaper and higher quality than Gemini 3.5 Flash on long-horizon tasks, and that higher thinking mode actually increases hallucination.
@fly3nn: Gemini is genuinely badass. I declare Gemini the most badass model in the world. I asked it to scrape WeChat public acc…
A user compares Gemini and GPT in scraping tasks, praising Gemini's advanced methods and criticizing GPT's less efficient approach, highlighting a significant performance gap.
The "One-Size-Fits-All" AI era is dead. I benchmarked GPT-5.5, Claude 4.7, Gemini 3.1 Pro, and DeepSeek V4 Pro here is the actual state of the frontier.
A benchmarking analysis of GPT-5.5, Claude Opus 4.7, Gemini 3.1 Pro, and DeepSeek V4 Pro reveals that no single model dominates all tasks; optimal performance requires a multi-model router with specialized model usage based on strengths and weaknesses.
Gemini 4 Carbon: Our next frontier model
Google DeepMind 发布 Gemini 4 Carbon，这是其最新的旗舰前沿模型，在网络安全（SEC-bench Pro、ExploitGym、CWE-bench）和长时程智能体任务上表现领先，目前通过 Fairwind 计划向可信网络安全防御者限量开放。
@interjc: Gemini 3.5 Pro 快着点儿吧，我都要把你开除出御三家了
据泄露的基准测试结果，Gemini 3.5 Pro 在内部评估中性能超过 Claude Fable 5 和 GPT-5.6，零样本性能相比 3.1 Pro 有显著提升，目前处于私下验证测试阶段，即将公开推出。