Latest news with #o1Pro
Business Times
5 days ago
- Business
- Business Times
Outperformed by AI: Time to replace your analyst?
SIX artificial intelligence (AI) models recently went head-to-head with seasoned equity analysts to produce Swot (strengths, weaknesses, opportunities and threats) analyses, and the results were striking. In many cases, the AI didn't just hold its own; it uncovered risks and strategic gaps the human experts missed. This wasn't theory. My colleagues and I ran a controlled test of leading large language models (LLMs) against analyst consensus on three companies – Deutsche Telekom (Germany), Daiichi Sankyo (Japan), and Kirby Corporation (the US). Each was the most positively rated stock in its region as of February 2025 – the kind of 'sure bet' that analysts overwhelmingly endorse. We deliberately chose market favourites because if AI can identify weaknesses where humans see only strengths, that's a powerful signal. It suggests that AI has the potential not just to support analyst workflows, but to challenge consensus thinking and possibly change the way investment research gets done. The uncomfortable truth about AI performance Here's what should make you sit up: With sophisticated prompting, certain LLMs exceeded human analysts in specificity and depth of analysis. Let that sink in. The machines produced more detailed, comprehensive Swots than professionals who have spent years in the industry. But before you eliminate the need for human analysts, there's a crucial caveat. While AI excels at data synthesis and pattern recognition, it can't read a CEO's body language or detect the subtext in management's 'cautiously optimistic' guidance. As one portfolio manager told us, 'Nothing replaces talking to management to understand how they really think about their business.' The 40 per cent difference that changes everything The most striking finding? Advanced prompting improved AI performance by up to 40 per cent. The difference between asking 'Give me a Swot for Deutsche Telekom' and providing detailed instructions is the difference between a Wikipedia summary and institutional-grade research. This isn't optional anymore — prompt engineering is becoming as essential as Excel was in the 2000s. Investment professionals who master this skill will extract exponentially more value from AI tools. Those who don't will watch competitors produce superior analysis in a fraction of the time. BT in your inbox Start and end each day with the latest news stories and analyses delivered straight to your inbox. Sign Up Sign Up The model hierarchy: Not all AI is created equal We tested and ranked six state-of-the-art models: 1. Google's Gemini Advanced 2.5 (Deep Research mode) – The clear winner 2. OpenAI's o1 Pro – Close second with exceptional reasoning 3. ChatGPT 4.5 – Solid but notably behind the leaders 4. Grok 3 – Elon Musk's challenger showing promise 5. DeepSeek R1 – China's dark horse, fast but less refined 6. ChatGPT 4o – The baseline for comparison The reasoning-optimised models (those with 'deep research' capabilities) consistently outperformed standard versions such as ChatGPT-4o. They provided more context, better fact-checking, and fewer generic statements. Think of it as hiring a senior analyst versus a junior analyst – both can do the job, but one needs far less handholding. Timing matters too. The best models took 10 to 15 minutes to produce comprehensive Swots, while simpler models delivered in less than a minute. There's a direct correlation between thinking time and output quality – something human analysts have always known. The European AI deficit: A strategic vulnerability Here's an uncomfortable reality for European readers: Of the models tested, five are American and one is Chinese. Europe's absence from the AI leadership board isn't just embarrassing – it's strategically dangerous. When DeepSeek emerged from China with competitive performance at a fraction of Western costs, it triggered what some called a 'Sputnik moment' for AI. The message was clear: AI leadership can shift rapidly, and those without domestic capabilities risk technological dependence. For European fund managers, this means relying on foreign AI for critical analysis. Do these models truly understand European Central Bank communications or German regulatory filings as well as they grasp US Federal Reserve statements? The jury's out, but the risk is real. The practical integration playbook Our research points to a clear four-step approach for how investment professionals should use these tools: 1. Hybrid, not replacement: Use AI for the heavy lifting – initial research, data synthesis, pattern identification. Reserve human judgment for interpretation, strategy, and anything requiring genuine insight into management thinking. The optimal workflow: AI drafts, humans refine. 2. Prompt libraries are your new alpha source: Develop standardised prompts for common tasks. A well-crafted Swot prompt is intellectual property. Share best practices internally, but guard your best prompts like trading strategies. 3. Model selection matters: For deep analysis, pay for reasoning-optimised models. For quick summaries, standard models suffice. Using GPT 4o for complex analysis is like bringing a knife to a gunfight. 4. Continuous evaluation: New models launch almost weekly. Our six-criteria evaluation framework (structure, plausibility, specificity, depth, cross-checking, meta-evaluation) provides a consistent way to assess whether the latest model truly improves on its predecessors. Beyond Swot: The expanding frontier While we focused on Swot analysis, the implications extend across the entire investment process. We list a few of these below, but there are many more: Earnings call summarisation and analysis in minutes, not hours ESG red flag identification across entire portfolios Regulatory filing analysis at scale Competitive intelligence gathering Market sentiment synthesis Each application frees human analysts for higher-value work. The question isn't whether to adopt AI – it's how quickly you can integrate it effectively. The uncomfortable questions Let's address what many are thinking: 'Will AI replace analysts?' Not entirely, but it will replace analysts who don't use AI. The combination of human plus AI will outperform either alone. 'Can I trust AI output?' Trust but verify. AI can hallucinate facts or miss context. Human oversight remains essential, especially for investment decisions. 'Which model should I use?' Start with Gemini Advanced 2.5 or o1 Pro (or the successors) for complex analysis. But given the pace of change, reassess quarterly. 'What if my competitors use AI better?' Then you'll be playing catch-up while they're finding alpha. Staying on the sidelines while competitors build AI advantage means ceding ground in an increasingly competitive landscape. The path forward The genie is out of the bottle. LLMs have demonstrated they can perform analytical work in seconds that once took days. They bring speed, consistency, and vast knowledge bases. Used effectively, they're like having a tireless team of junior analysts who never sleep. But here's the key: Success requires thoughtful integration, not wholesale adoption. Treat AI output as you would a junior analyst's draft – valuable input requiring senior review. Master prompt engineering. Choose models wisely. Maintain human oversight. For European professionals, there's an additional imperative: Push for domestic AI development. Technological dependence in critical financial infrastructure is a strategic vulnerability no region can afford. Master the tools – or be outpaced by them Embrace these tools intelligently, or watch competitors leave you behind. The winners in this new landscape will be those who combine AI's computational power with human insight, intuition, and relationship skills. The future of investment analysis isn't human or AI – it's human and AI. Those who recognise this and act accordingly will thrive. Those who don't will find themselves outperformed not by machines, but by humans who learned to work with them. Your next analyst hire might still need that coffee break. But they'd better know how to prompt an LLM, evaluate its output, and add the human insight that transforms data into alpha. Because in 2025, that's the new standard. The tools are here. The frameworks exist. The winners will be the ones who know how to use them. This content has been adapted from an article that first appeared in Enterprising Investor at The full study can be found here The writer, CFA, is chief investment officer at MHS CapInvest, where he employs advanced AI tools to enhance allocation, stock selection, portfolio construction, and risk management for different market capitalisations, He trains teams at DAX-listed companies on generative AI integration and helps investment professionals leverage tools like ChatGPT and Gemini to enhance their performance.
Hans India
12-06-2025
- Business
- Hans India
OpenAI Launches o3 Pro, Slashes o3 Model Costs by 80%, Delays Open-Source Model Release
In a major update for developers and AI users, OpenAI has rolled out its new o3 Pro model to ChatGPT Pro and Team users, marking its most advanced AI release to date. At the same time, the company has significantly lowered the cost of its standard o3 model by 80%, further widening accessibility. However, plans for releasing an open-source AI model have been pushed to later this summer. The o3 Pro model, now available through ChatGPT and OpenAI's API, replaces the older o1 Pro and brings a higher level of reasoning and accuracy, especially in areas like science, education, programming, and mathematics. OpenAI describes it as 'a version of our most intelligent model, o3, designed to think longer and provide the most reliable responses.' According to the company, the new model has shown superior performance in both internal and academic evaluations, particularly in clarity, instruction adherence, and depth of analysis. What sets o3 Pro apart is its enhanced reliability, a feature that makes it ideal for handling complex queries where precision is critical. As OpenAI notes, 'We recommend using it for challenging questions where reliability matters more than speed, and waiting a few minutes is worth the tradeoff.' Despite using the same underlying architecture as o3, the Pro version has been optimised for dependability. It includes advanced tools like Python code execution, document analysis, web browsing, visual input interpretation, and memory-based personalisation. These tools make o3 Pro more versatile, though response times are typically longer compared to o1 Pro. Notably, some capabilities like temporary chats, image generation, and the Canvas interface are not yet available in o3 Pro. OpenAI has advised users to stick with GPT-4o, o3, or o4-mini for those particular features. Enterprise and Education customers will gain access to the model in the upcoming week. In tandem with the model upgrade, OpenAI announced a dramatic cost reduction for its o3 model—from $10 to $2 per million input tokens and from $40 to $8 per million output tokens. Cached prompt usage comes with further discounts. The update places OpenAI in a more competitive pricing bracket compared to rivals like Google DeepMind's Gemini and Anthropic's Claude. Confirming the change, CEO Sam Altman posted on X: 'we dropped the price of o3 by 80%!! excited to see what people will do with it now. think you'll also be happy with o3-pro pricing for the performance :)' However, not all announcements were forward-moving. OpenAI's open-source AI model, initially expected in June, has been delayed. Altman shared that the postponement is due to unexpected progress by the research team that requires additional refinement. 'We are going to take a little more time with our open-weights model, i.e. expect it later this summer but not June,' he wrote. The open-source model is anticipated to compete with the likes of DeepSeek R1 and is designed to raise the standard for freely accessible large language models. In a separate blog post, Altman also addressed environmental concerns around AI use, revealing that a single ChatGPT query consumes approximately 0.34 watt-hours of electricity and about 0.000085 gallons of water—comparable to a second of oven use or a sip of water. He added, 'The cost of intelligence should eventually converge to near the cost of electricity.' With these moves, OpenAI not only aims to empower developers and enterprises with powerful tools at lower costs but also continues to push the envelope in AI innovation.
India Today
11-06-2025
- Business
- India Today
OpenAI adds o3 Pro to ChatGPT and drops o3 price by 80 per cent, but open-source AI is delayed
OpenAI has been busy. This morning, we woke up to a bunch of updates from the AI company. To begin with, OpenAI has released o3 Pro, its most advanced reasoning model to date, for Pro and Team users of ChatGPT. The release of the new model also comes with a dramatic 80 per cent price cut for the standard o3 model, which makes it more accessible for developers and researchers around the world. However, OpenAI CEO Sam Altman has also announced that the company has had to push back the release of its open-source model until later this Pro lands in ChatGPTOpenAI has launched a new o3 Pro model, which it says is its most capable AI model yet. o3 Pro is available to Pro and Team users in ChatGPT and through OpenAI's API. It replaces the earlier o1 Pro model and is designed to deliver even more reliable and accurate answers across complex domains like science, programming, mathematics, and education. OpenAI described o3 Pro as a 'version of our most intelligent model, o3, designed to think longer and provide the most reliable responses.' The company notes that in internal and external academic evaluation, o3 Pro has consistently outperformed both o3 and its predecessors, especially in clarity, accuracy, instruction-following, and comprehensiveness. advertisement Although o3 Pro uses the same underlying model as o3, it is optimised for dependability. OpenAI says reviewers have ranked the model higher across key tasks in science and data analysis. As OpenAI puts it, 'We recommend using it for challenging questions where reliability matters more than speed, and waiting a few minutes is worth the tradeoff.' Notably, o3 Pro is also equipped with tools that enhance its capabilities, including the ability to browse the web, analyse documents, interpret visual inputs, run Python code, and use memory for more personalised interactions. However, these advanced tools mean its responses typically take longer to generate than those of o1 the rollout, a few features remain unavailable in o3 Pro for now. Temporary chats are currently disabled due to a technical issue, and the model cannot yet generate images or use the Canvas interface. Users looking for image generation are advised to stick with GPT-4o, o3, or and Edu customers will gain access to o3 Pro the following week. A massive price cut for o3Alongside the launch of o3 Pro, OpenAI has also slashed the cost of using the o3 model by 80 per cent – a move that makes the model more accessible and affordable to developers. The cost of using o3 now stands at $2 per million input tokens and $8 per million output tokens, with additional discounts for cached (previously seen) prompts. This is a major reduction from the previous rates of $10 and $40 respectively. As OpenAI researcher Noam Brown pointed out on X, this update represents a significant shift in affordability for CEO Sam Altman confirmed the price drop on X, posting, 'we dropped the price of o3 by 80%!! excited to see what people will do with it now. think you'll also be happy with o3-pro pricing for the performance :)' advertisementThe new pricing places OpenAI in sharper competition with rivals like Google DeepMind's Gemini 2.5 Pro, which charges between $1.25 and $2.50 for input and up to $15 for output tokens, depending on prompt size. Gemini also offers Google Search integration, but with extra fees beyond a certain usage Claude Opus 4 remains the priciest competitor, charging $15 per million input tokens and $75 for output. DeepSeek's Reasoner and Chat models, on the other hand, lead the low-cost tier with rates as low as $0.035 for cached inputs during off-peak model postponedWhile the launch of o3 Pro and the price drop for the o3 model are widely welcomed, OpenAI's open-source model has hit a delay. Originally expected in June 2025, the model will now arrive later, following a surprise development by the research team. The exact timeline for the release has not yet been disclosed. In a post on X, Altman explained, 'We are going to take a little more time with our open-weights model, i.e. expect it later this summer but not June.' He added that the delay is due to the team achieving something 'unexpected and quite amazing,' which still needs more time to open-source model is expected to rival high-performing open reasoning models such as DeepSeek R1, with OpenAI aiming to raise the bar for open-access large language models. OpenAI CEO reveals cost of a single ChatGPT queryAlongside these product updates, Altman has also addressed questions about the environmental cost of using ChatGPT. In a blog post published on Tuesday, he shared that each ChatGPT query consumes about 0.34 watt-hours, which is roughly what an oven uses in a second or a high-efficiency lightbulb in a couple of terms of water usage, Altman revealed that the average query consumes 0.000085 gallons of water, or 'roughly one fifteenth of a teaspoon.' These figures, he explained, are part of a broader vision where 'the cost of intelligence should eventually converge to near the cost of electricity.'
