I've been using GPT-5, and it's a major 昇格 ChatGPT 使用者s don’t want
When you buy through links on our articles, 未来 and its syndication partners may earn a （売買）手数料，委託（する）／委員会／権限.
AI Byte
AI Byte is a 週刊誌 column covering all things 人工的な 知能, 含むing AI models, apps, features, and how they all 衝撃 your favorite 装置s.
OpenAI has the （疑いを）晴らす brand advantage in the AI race, and that's something that may 証明する impossible for competitors to 打ち勝つ. ChatGPT has more 週刊誌 使用者s than Meta or Google has on a 月毎の basis. 存在 first comes with 利益s, which OpenAI is certainly 得るing.
In the time since, though, the 競争 caught up ― at least on a 機能の level. AI models from Google, DeepSeek, Claude, and xAI have all held 最高の，を越す-five 位置／汚点／見つけ出すs on LMArena's leaderboard, signaling that the race from basic large language models (LLMs) to 人工的な general 知能 (AGI) is far from decided. For months, I've been 確かな that Gemini 2.5 プロの／賛成の is a better and more versatile thinking model than anything coming out of OpenAI.
井戸／弁護士席, OpenAI is hoping to change that narrative with the 解放（する） of GPT-5, which 開始する，打ち上げるd for everyone last week (Aug. 7). It's 現実に a collection of models that can intelligently 選ぶ which one is best for a given 誘発する on the 飛行機で行く. And at least for now, it's at the 最高の，を越す of both the LMArena and WebDev 円形競技場 leaderboards ― two 決定的な LLM （判断の）基準s.
GPT-5 doesn't solve all of ChatGPT's hallucination problems, and it's definitely not AGI. There are some areas where Google models still outperform 類似の ones from OpenAI. Regardless, GPT-5 looks impressive and 解放する／自由な, and that might be all OpenAI needs to 持続する its hefty lead over the 競争.
Where OpenAI's GPT-5 （警官の）巡回区域，受持ち区域s Google's Gemini 2.5 プロの／賛成の
GPT-5’s smartest feature has nothing to do with 計算する or its knowledge base. It’s called real-time 大勝するing, and it helps 選ぶ the 権利 model for your 仕事 without any 付加 使用者 input. 現在／一般に, AI chatbots like ChatGPT and Gemini are littered with a variety of new, old, and 実験の models best ふさわしい for a 明確な／細部 誘発する. That’s 広大な／多数の／重要な, but the onus was on the 使用者 to decide which one is best. With 指名するs like GPT-o3 and GPT-4o, or Gemini Flash Thinking 実験の, making that 決定／判定勝ち（する） wasn’t always 平易な.
Instead, GPT-5 is a simple 指名する for a collection of OpenAI models. It 含むs a lightweight model for quick, 平易な 誘発するs and a more thoughtful model called GPT-5 Thinking for コンビナート／複合体 queries. OpenAI’s real-time router means that all of these models 機能（する）／行事 as one, at least to the 使用者. After inputting a 誘発する to ChatGPT, the GPT-5 router will decide which one to use, 簡素化するing the UX.
There are still ways to manually 支配（する）／統制する which GPT-5 model 答える／応じるs to your 誘発する. You can use phrases like “think really hard about this” in your 誘発する to 誘発する／引き起こす GPT-5 Thinking. If ChatGPT 誤った 選ぶs for the thinking model, 使用者s can tap Get a quick answer to pivot to the lightweight model instead. Both models appear to be very reliable in 早期に 実験（する）ing, 明確に 特記する／引用するing sources to 避ける hallucinations.
These online sources, which 含む 場所／位置s like Android Central, are a 重要な 推論する／理由 why hallucinations are 負かす／撃墜する with GPT-5. OpenAI says that with web search enabled on GPT-5 誘発するs, the model is about 45% いっそう少なく likely to 含む／封じ込める a factual error than GPT-4o. Hallucinations are by no means extinct, but they are いっそう少なく 流布している when using GPT-5 in ChatGPT.
独立した・無所属 実験（する）s and （判断の）基準s seem to 提携させる with OpenAI’s （人命などを）奪う，主張するs that GPT-5 is better at 令状ing, coding, and health-based 仕事s. It finally usurps Gemini 2.5 プロの／賛成の on both the LMArena and WebDev 円形競技場 leaderboards, （人命などを）奪う，主張するing the 最高の，を越す 位置／汚点／見つけ出す 全体にわたる. 特に, GPT-5 seems to have the 辛勝する／優位 in text-based and coding 誘発するs. I used a 見本 誘発する from OpenAI to 実験（する） GPT-5’s coding 能力s in ChatGPT, and （機の）カム away 本気で impressed.
Where Gemini still （警官の）巡回区域，受持ち区域s ChatGPT
GPT-5 isn’t better than Gemini 2.5 プロの／賛成の in every area, and OpenAI still doesn’t seem to be able to match Google’s image and ビデオ 世代. That’s 反映するd in LMArena’s text-to-image, text-to-ビデオ, and image-to-ビデオ （判断の）基準s ― Google’s Imagen 4 and Veo 3 道具s sweep the graphical 控訴 of generative AI 実験（する）s.
I 手配中の，お尋ね者 to 実験（する） that 結果 in the real world, so I fed ChatGPT and Gemini the same 誘発する: "生成する an image of Johnny Thunderbird 持つ／拘留するing up a Big East Tournament トロフィー at Madison Square Garden." Here’s how the images turned out:
Anecdotally, the 生成するd images 提携させる with the （判断の）基準 results ― Google’s ビデオ- and image-世代 chops are much stronger than OpenAI’s. For starters, Google gave me a 生成するd image in about 10 seconds, and I had to wait nearly two minutes for an image to come 支援する from ChatGPT.
From an 正確 見地, Gemini won handedly. It knew I was talking about my 母校’s basketball program from the mascot 言及／関連, and 正確に recreated Madison Square Garden with a トン of 詳細（に述べる） and flair. 一方／合間, ChatGPT produced an image with the wrong team and I’m not 納得させるd the generic background is really MSG; it looks like it could be any basketball 法廷，裁判所.
So, if you’re the 肉親，親類d of person who likes to switch between LLMs based on your 仕事, perhaps choose GPT-5 for 令状ing or coding and Gemini for image or ビデオ 世代.
GPT-5 is 広大な／多数の／重要な, but GPT-4o just won't go away
A wrinkle in OpenAI’s GPT-5 rollout is 使用者 attachment to older models. GPT-5 was more than two years in the making, as OpenAI 報道によれば struggled to make an AI model worthy of the 肩書を与える. As it turns out, 使用者s didn’t want one. OpenAI is bringing 支援する GPT-4o ― a model that was supposed to be sunset in 好意 of GPT-5 ― because of 使用者 （激しい）反発 (経由で Tom’s Guide).
It might seem silly for those of us who touch grass, but ChatGPT 使用者s have seemingly created parasocial 関係s with OpenAI models. They’d rather use the AI model they “know” over GPT-5, even if the newer one is better in every way.
I think you can extrapolate this 前提 to the broader AI race. In some ways, it really doesn’t 事柄 whether Google, OpenAI, Claude, or DeepSeek has the best-成し遂げるing model. People are going to stick to the models they like and are familiar with, and if that’s the 事例／患者, OpenAI’s lead in this space may be insurmountable for the 競争.