Microsoft launches AI Accelerator for Sales, AI agents near public previews

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

Microsoft launched a series of AI agents and an AI accelerator program for sales as it outlined a bevy of customer references implementing the software giant's copilots and agents.

The company's AI agent tools are being rolled out in Dynamics 365 and are natively integrated across Microsoft 365 productivity apps. The AI agents will connect to an enterprise's data platform (Azure and third parties) and use Microsoft's horizontal data across productivity apps and LinkedIn to include context.

Here's what's rolling out:

• AI Accelerator for Sales, a program that brings services and expertise to enterprises adopting agentic AI, which will be available in April.
• Sales Research Agent, which will be in public preview in April.
• Sales Agent, which brings agentic AI to salespeople in various environments such as Dynamics 365 and Outlook. Sales Agent will be in public preview in May and autonomously researches and prioritizes leads.
• Sales Chat, which allows you to interact with sales data in natural language, will public preview in May.

Sales Agent and Sales Chat agents can be accessed in Microsoft 365 Copilot, Copilot Chat, Dynamics 365 and Salesforce.

Microsoft also announced a variety of customer references for Microsoft 365 Copilot. Many of these customers were highlighted by CEO Satya Nadella on Microsoft's most recent earnings conference call.

• Microsoft launches Copilot Actions, Agents in SharePoint at Ignite
• Microsoft outlines Copilot agents for enterprises

The list of customers hits back at Salesforce CEO Marc Benioff who has argued repeatedly that Microsoft couldn't show what customers were doing with Copilot or AI agents. Microsoft said more than 160,000 organizations have used Copilot Studio to create more than 400,000 customer agents in the last three months.

• AI agents bring consumption models to SaaS: Goldilocks or headache?
• Agentic AI: Three themes to watch for 2025
• Agentic AI without process optimization, orchestration will flop

Enterprise customers leveraging AI agents and Copilot include Holland America Line, Vodafone, Novartis, Campari, Dow, ABN Amro, Accenture, and Estee Lauder to name a few.

Constellation Research got a demo of Sales Agent, Sales Chat and an AI canvas applications that enabled enterprises to delve into data and dashboards in natural language. A few takeaways:

• The ability for Sales Agent to interact with sales folks within various applications like Outlook with context in an autonomous way was notable. Tasks can be sped up dramatically.
• Sales Agent can carry out tasks and automatically update Dynamics 365.
• Performance tracking on engagements on metrics such as emails sent and open rates, meetings booked and opportunities created enable enterprises to gauge ROI of agents and optimized.
• A Sales Management AI workspace highlights the future of workflows as well as the ability to get specific data points and discover new insight. This "infinite canvas" approach can be helpful and enable enterprises to make better decisions.

Constellation Research analyst Martin Schneider outlined his thoughts in a post. He said:

"These advancements underscore an overarching theme occurring across front-office application development. Where we used to think of suites like CRM as a collections of applications with their own user interfaces….today we are seeing “Copilots” (and their relative cognates) as the new collective UI that can draw functionality and present insights from a wide swath of applications. Also, where we saw workflow as something that needed to be considered, built, and managed - now agents have streamlined workflow to the point where we can simply create complex and effective workflows with a single sentence added to a prompt. And where we looked at CRM systems as a data store, now we can finally start looking at them as a tool for access to knowledge beyond just the data we put into it. High composability, the simplicity of conversational interfaces and more fluid data integration capabilities are driving these rapid changes."

CrowdStrike delivers strong Q4, but earnings outlook light

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

CrowdStrike reported better-than-expected fourth quarter results as the company was seeing strong enterprise demand for its AI driven security platform

The company, which is competing with Palo Alto Networks as a "platformization" play, reported a fourth quarter net loss of $92.3 million, or 37 cents a share, on revenue of $1.06 billion, up 255 from a year ago. Non-GAAP fourth quarter earnings were $1.03 a share.

Wall Street was expecting CrowdStrike to report non-GAAP earnings of 86 cents a share on revenue of $1.04 billion.

For fiscal 2025, CrowdStrike reported a net loss of $19.3 million, or 8 cents a share, on revenue of $3.95 billion, up 29% from a year ago. The company appears to be fully recovered from its July outage. 

• Cybersecurity platformization: What you need to know
• CrowdStrike, Palo Alto Networks duel over platforms vs. bundles
• Palo Alto Networks delivers strong Q1, says industry on platformization bandwagon now
• How Platformization Applies to Cybersecurity

CrowdStrike CEO George Kurtz said the company is "on the flight path to our $10 billion ending ARR (annual recurring revenue) goal." Kurtz added the company was seeing strong demand for its Next-Gen SIEM, Cloud Security, and Identity Protection businesses. Falcon Flex, CrowdStrike's cybersecurity platform, added more than $1 billion in deal value in the fourth quarter.

As for the outlook, CrowdStrike projected first quarter revenue of $1.1 billion or so with non-GAAP earnings of 64 cents a share to 66 cents a share. Wall Street was expecting non-GAAP earnings for the first quarter of 96 cents a share. For fiscal 2026, CrowdStrike projected revenue of $4.74 billion to $4.8 billion with non-GAAP earnings of $3.33 a share to $3.45 a share. Analysts were expecting $4.43 a share for fiscal 2026. 

• CrowdStrike buys Adaptive Security
• CrowdStrike, AWS expand partnership revolving around CrowdStrike Falcon and Amazon Bedrock

Speaking on a conference call, Kurtz said:

• "CrowdStrike is the first cybersecurity ISV to cross $1 billion in deal value on AWS Marketplace in a single calendar year, setting a new standard for ecosystem execution."
• "We find ourselves placed at the epicenter of a rapidly evolving demand environment. A new administration, a new wave of technology, and a new threat landscape necessitate all businesses to evolve their cybersecurity programs. Consolidation, cost reduction and automation are now the accepted enterprise and federal priorities."
• "We're still in the early, but rapidly evolving innings of the AI revolution. Businesses and governments across the globe are looking for their AI investments to yield both improved efficiencies and novel outputs. At CrowdStrike, we're requiring every team and function to leverage the power of AI."
• "Businesses are equally grappling with how to secure their environments in this AI age. Here's my take on what this means. First, more AI everywhere means more data, more access, and more processes, services, and products requiring cybersecurity."
• "With tools such as DeepSeek making AI access easier and cheaper, the pace and prevalence of adversarial AI adoption is only accelerating."
• "More access to more third-party and in-house agentic applications and services requires rethinking identity and data protection. Who is accessing data and where is it traveling matters more now than ever before." 
• "Securing AI starts a broader enterprise data discussion. I'm seeing CISOs, CIOs, and CEOs going to the drawing board to reinvent their technology stack with AI-powered platforms of record for their next decade and beyond. And for security, it's even more pressing."

AI agents aimed at healthcare, but data interoperability still a big issue

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

Google Cloud and Microsoft are launching generative AI tools aimed at healthcare by streamlining documentation and searching unstructured data to optimize workflows. Vendors keep rolling out healthcare focused AI and automation, but the industry still struggles with transformation as well as data interoperability.

Google Cloud said it added a feature called Visual Q&A in its Vertex AI Search for healthcare. Visual Q&A is designed to search tables, charts, medical images and diagrams for data. Google Cloud also said Gemini 2.0 models are available in its healthcare offering.

The idea is that physicians will leverage multimodal search and models to get a better view of patient health.

Microsoft launched Microsoft Dragon Copilot, an AI assistant for clinical workflows. The Dragon Copilot will be combined with DAX, a natural language voice dictation platform that includes ambient listening, genAI models and healthcare guardrails. Microsoft acquired Nuance in 2022 for its healthcare foothold and later launched Microsoft Cloud for Healthcare.

According to Microsoft, Dragon Copilot can streamline documentation with multilanguage ambient note creation, surface medical information from trusted sources and automate tasks such as notes, evidence summaries and referral letters. Microsoft's Dragon Copilot can also navigate electronic health records.

The news from Microsoft and Google Cloud are a barrage of releases timed for HIMSS, a large healthcare technology conference. 

• Oracle launched its Oracle Health Clinical AI Agent, which is a multimodal and screen driven assistant, is available for more than 30 medical specialties. The company said physicians using the AI agent are seeing a decrease of 30% in documentation time. 
• Salesforce made a similar claim as it launched Agentforce for Health, a library of prebuilt skills and actions for healthcare teams.
• Epic, which is essentially the ERP system for healthcare, also said it added a host of generative AI tools to make healthcare workflows more efficient.  

AI in healthcare: Are we there yet?

At Constellation Research's Ambient Experience Summit 2025, there were multiple talks about healthcare, AI and patient experiences. See: Constellation Research Ambient Experience Summit 2025: 10 CX takeaways on people, data, AI

Joseph Anzalone, VP of Marketing at GE Healthcare, said the company is transforming from a capital equipment company to one focused on AI applications. GE Healthcare is planning to leverage the 96% of unused data from its devices to enhance clinician experiences.

"We're focusing on the clinician experience now and building AI applications on top of foundation models that will give clinicians usable data and predictive analytics," said Anzalone, who added that healthcare workflows are often slowed by data silos across various systems, insurance providers and channels.

• How GE Healthcare is approaching generative AI, LLMs, and transformation

The bet is that AI can help overcome legacy infrastructure.

Kimberly Powell, VP of Healthcare at Nvidia, said digital devices, digital biology and clinical will all be big markets for agentic AI systems.

Powell argued that AI agents will serve as a digital workforce due to shortages. AI agents will succeed because they can sit on top of existing infrastructure. She said:

"I believe agentic AI can overlay on the healthcare system unlike technology in the past. Agents can connect and have a foundation model through an API. Since there's an API, AI agents can navigate autonomously over what is otherwise a antiquated and very complex and disaggregated healthcare system."

Also see: BT150 Spotlight: Centurion Health's Johnny Wu on delivering healthcare in prisons

• BT150 Spotlight: MultiCare Health System's Laurie Wheeler on optimization, change management and healthcare transformation
• BioNtech, InstaDeep bet on genAI models to advance R&D, drug discovery, cancer treatment
• AI 15O's Matt Lewis on GenAI adoption, psychology and life sciences
• Welldoc Chief AI Officer Anand Iyer on AI, data and healthcare transformation
• How ResMed’s data prowess sets it up for AI, sleep health market expansion

Data interoperability needed

While these AI visions for healthcare are in progress and impressive, Constellation Research CEO Ray Wang argued in a report that the industry needs to focus on healthcare data interoperability.

Wang said:

"Healthcare systems around the world are facing a confluence of challenges such as aging and declining populations, rising cost of delivery, clinical workforce shortages, and increasing expectations for better outcomes at lower prices. Although healthcare systems have accelerated their digitization projects in an effort to modernize, the results have not shown massive efficiency. One of the biggest opportunities to accelerate digitization and prepare for the Age of AI is improved healthcare data interoperability. In fact, interoperability will play a significant role in driving down costs and improving patient outcomes."

The industry needs to focus on interoperability because "current patients, physicians, and healthcare providers and payors face a constant struggle to integrate, manage, and orchestrate across a hodgepodge of physical and digital functional fiefdoms and silos."

Ultimately, the healthcare industry needs to focus on vendors that support data interoperability, hone data consistency, make interoperability part of governance and leverage standards.

Qualcomm fleshes out Dragonwing with fixed wireless access platform

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

Qualcomm fleshed out its Dragonwing platform as it launched a series of fixed wireless access building blocks designed to enable AI inference use cases as well as industrial applications.

At Mobile World Congress, Qualcomm aimed to stretch beyond its 5G modems, smartphones and PCs to enterprise use cases. Qualcomm launched its Dragonwing branding designed to target enterprises and industrial IoT use cases as well as a partnership with IBM.

The company is looking to expand from handsets, autos, PCs and augmented reality to industrial, networking and other devices.

Qualcomm launched Dragonwing Fixed Wireless Access Gen 4 Elite, a platform that has 40 TOPS Edge AI integration, tops downlink speeds up to 12.5 Gbps and has on-device AI-enhanced traffic classification.

• Qualcomm launches Dragonwing, expands into industrial AI
• Qualcomm Q1 strong as on-device generative AI ramps
• Top 5 Takeaways from Qualcomm AI Day | With Analyst Andy Thurai
• Qualcomm, Arm cheer cheaper models, AI inference at the edge

To go along with Dragonwing, Qualcomm launched X85 5G Modem-RF. The Dragonwing FWA Gen 4 Elite platform can deliver long-range 5G mmWave, NTN Satellite communication and dual SIM connectivity along with cellular, Wi-Fi and Ethernet.

The news from Qualcomm is part of a broader strategy to be seen as an AI processor company as more workloads go to the edge of the network. The company launched a series of modems--Qualcomm E41 4G Modem-RF, Qualcomm E51 4G Modem-RF and Qualcomm E52 4G Modem-RF--designed for industrial use cases.

According to Qualcomm, the modems are designed to be deployed in multiple environments such as warehouses, manufacturing plants and other areas with poor signals. The modems can be positioned in any environment and incorporates iSIM technology into the hardware for easy programming.

With the effort, Qualcomm is looking to develop use cases in energy and utilities, retail and autonomous robots.

• Physical AI, world foundation models will move to forefront

Will CX suffer from an overreliance on AI?

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

Although AI--generative, agentic and everything in between--is viewed as a tool that can improve customer experience there is the risk that enterprises may rely too much on the technologies.

At Constellation Research's Ambient Experience Summit 2025, we held a workshop on ways to think different about experiences and address one key challenge. Over-reliance on AI was a big theme as was disjointed customer journeys.

• Constellation Research Ambient Experience Summit 2025: 10 CX takeaways on people, data, AI
• AX100 Spotlight: RentSpree's Michael Storey on the intersection of CX, AI and success metrics

One group noted that CX can be harmed by overreliance on AI. AI can only be so personal and the big workflow issue is figuring out where to insert a human into processes. To date, AI is viewed as a productivity enhancement that can add digital labor and in many cases replace humans.

Yes, CX leaders need to implement AI, but have no control about how it's deployed. In addition, the models can be experimental and simply off the mark. AI isn't a human replacement and it's a big reason why we're hitting #0 every time we call customer service.

Constellation Research analyst Liz Miller noted:

"Let's be honest, if we keep training AI on the nonsense that it's been trained on so far, it is going to get even worse. I do not want my customer experience to be over reliant on AI, because it can be unreliable, lacks human touch, and it is experimental, but it can help when it's deployed."

A few takeaways from the workshop:

• CX leaders need to think through AI's deficiencies and know the problems they're trying to solve. Is the issue cost efficiency? Is it growth? Is it boosting the lifetime value of the customer relationship?
• CX leaders need to think through the customer journey and where AI can help. Why are journeys mapped the way they are? It's quite possible that AI and workflow automation may make the customer journey even more disjointed.
• AI is going to occupy a larger chunk of the customer experience, but don’t go crazy. Humans will have to find the balance needed to blend AI with real-life touchpoints.

One company worth watching on this AI-CX front is Intuit. Speaking on Intuit's fiscal second quarter earnings conference call, CEO Sasan Goodarzi talked about how it was blending AI and humans to create experiences with Intuit Assist.

"We are making strong progress across our platform with our data and AI investments to deliver done-for-you experiences with AI-powered human expertise,” said Goodarzi. “Our focus is on automating tasks, end-to-end workflows, and entire functions, connecting customers to one of our more than 12,000 AI-powered human experts for that last mile or to complete all of the work."

Goodarzi said Intuit is seeing higher conversions across its product line with personalized product experiences. "With the scale of our data and AI capabilities, Intuit Assist is the control tower automating tasks and workflows, with human experts engaging where needed to deliver done-for-you experiences for our customers," said Goodarzi.

Intuit also highlights one of the key takeaways from our CX summit: Data quality is everything. Intuit has invested heavily in its cloud-based data architecture that set the stage for AI.

• Intuit embraces LLM choice for multiple use cases
• Intuit sees 'green shoots' from its generative AI strategy
• Intuit’s bets on data, AI, AWS pay off ahead of generative AI transformation (PDF)

Dell Q4 powered by AI servers

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

Dell Technologies delivered better-than-expected fourth quarter earnings as it continues to see strong demand for AI servers. Shipments for AI servers will hit $15 billion in fiscal 2026, according to the company.

The company reported fourth quarter earnings of $2.15 a share on revenue of $23.9 billion, up 7% from a year ago. Non-GAAP earnings were $2.68 a share.

Wall Street was looking for earnings of $2.53 a share on revenue of $24.57 billion.

Dell Technologies' Infrastructure Solutions Group delivered revenue growth of 22%. Jeff Clarke, chief operating officer, said:

"Our prospects for AI are strong, as we extend AI from the largest cloud service providers, into the enterprise at-scale, and out to the edge with the PC. The deals we’ve booked with xAI and others puts our AI server backlog at roughly $9 billion as of today."

Dell also raised its annual dividend by 18%.

For fiscal 2025, Dell reported earnings of $6.38 a share on revenue of $95.6 billion, up 8% a year ago.

As for the outlook, Dell projected fiscal 2026 revenue growth of 8% to between $101 billion and $105 billion and non-GAAP earnings per share growth of 23%.

Speaking on an earnings conference call, Clarke said:

"We booked deals putting our AI backlog at roughly $9 billion as of today. Our pipeline expanded sequentially and has grown every quarter since the introduction of the 9680. We are seeing continued progress in AI from enterprise customers, albeit still earlier in their journey with sequential growth in both orders and customers. And our engineering services, financing and ability to optimize density and performance per watt are important differentiators for the largest at-scale CSPs and provide very efficient enterprise solutions. In traditional servers, the growth trajectory continues, up double digits in Q4. We've now seen 5 quarters of year-over-year demand."

As for the outlook, Dell CFO Yvonne McGill said:

"IT spending is expected to grow with 3 underlying trends that we see. First, businesses are leveraging AI to enable competitive advantages, and we are seeing that in our opportunity pipeline that continues to expand. Second, data center modernization is well underway with a focus on consolidation and power efficiency. Third, customers are planning to refresh their PC installed base with AI-enabled devices."

By the numbers:

• Infrastructure Solutions Group (ISG) fourth quarter revenue was $11.4 billion, up 22% from a year ago. Annual revenue was $43.6 billion, up 29%.
• Servers and networking revenue in the fourth quarter was $6.6 billion, up 37% due to AI servers.
• AI backlog was $4.1 billion exiting the fourth quarter.
• Storage revenue was up $4.7 billion, up 5%.
• Operating income for ISG in the fourth quarter was $2.1 billion.
• Dell's PC unit revenue in the fourth quarter was $11.9 billion, up 2% from a year ago, with operating income of $631 million. Commercial client revenue was up 5% in the fourth quarter and consumer revenue fell 12%.

Other Dell items worth noting:

US government exposure. "We've had numerous times in our history where a country or a particular segment demand was suppressed for various reasons. We've been able to navigate the cycles, I think, pretty successfully," said Clarke. "Our underlying belief is United States government will need technology. AI plays a pretty significant role in our nation. And I think the demand will materialize. We'll get through whatever is happening today."

Tariffs. Clarke said:

"This is a pretty darn dynamic environment as represented what we heard this morning. It's fluid. We built an industry-leading supply chain that's globally diverse, agile, resilient that helps us minimize the impacts of these trade regulations, tariffs to our customers and shareholders. We've been monitoring this for some time. We've taken our digital supply chain, our digital twins actually using some AI modelling to look at every possible scenario that you might imagine of this country, that country, and restrictions to help us understand how we optimize our network and how we that in the least amount of time at the speed of Dell. And whatever tariff we cannot mitigate, we view that as an input cost. And as our input costs go up, it may require us to adjust prices." 

Nvidia Blackwell margins. Clarke said Blackwell margins remain lower than Hopper. "We're still early. The deals are very large upfront. There's more competitors, so it's a more competitive landscape," said Clarke.  He added:

"This is system design and architecture work. There's an ability to really distinguish our engineering and value add in that step, which is an opportunity for us to extract value and opportunity for us to reduce cost. These aren't reference designs or as we would affectionately call in the engineering community, they're not cookie-cutter designs. We're designing a unique rack, a unique power distribution unit. Our cooling, our manifold, the cold plate, the ability to engineer that and to drive that through the scale of our supply chain are opportunities for us, helping our customers attach with our networking with our storage or opportunities."

Constellation Research analyst Holger Mueller said it remains to be seen how on-premises AI can boost demand. "Dell manages to beat inflation, but not by much, as its client solutions business is still dragging, with even the commercial business barely clocking in at inflation rate," he said. "If Dell’s ISG business keeps growing as it did in the last fiscal, ISG will pass CSG, a milestone for Dell. Dell became more profitable, with EPS up by almost 40%. AI demand keeps being the main driver. The big question is: How strong will the on-premises AI bonanza be in 2025?"

OpenAI's GPT-4.5 has more emotional intelligence, could expand use cases

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

OpenAI launched a research preview for GPT 4.5 and perhaps the biggest takeaway is that it can fake emotional intelligence pretty well--potentially better than a few humans we know.

The company outlined GPT-4.5 and noted that early testing shows that interactions with humans are more natural. OpenAI noted:

"Based on early testing, developers may find GPT‑4.5 particularly useful for applications that benefit from its higher emotional intelligence and creativity—such as writing help, communication, learning, coaching, and brainstorming. It also shows strong capabilities in agentic planning and execution, including multi-step coding workflows and complex task automation."

• OpenAI's Stargate Project diversifies its cloud infrastructure
• OpenAI launches GPT Store, Team plan, eyes developer, enterprise traction

For collaboration, GPT-4.5 may be a win since it has more EQ.

Key points about GPT-4.5:

• GPT-4.5 scales unsupervised learning by scaling up compute and data and reduces hallucinations.
• GPT-4.5 scored better than GPT-4o on collaboration skills.
• The model doesn't think before it responds.
• ChatGPT Pro users can select GPT-4.5 today with Pro and Team users getting it next week followed by Enterprise and education users.
• GPT 4.5 is compute intensive and more expensive than GPT-4o. It is not viewed as a replacement.

Constellation Research’s take

Andy Thurai, analyst at Constellation Research, said:

“OpenAI claims GPT-4.5 will have fewer hallucinations and more accurate writing, programming, and insights.

This is not a reasoning model but an unsupervised learning model. GPT-4.5 doesn’t have a chain of thought like the reasoning models therefore can respond faster. Instead of combining the reasoning models and responding models, OpenAI chose to separate them as the execution costs, and model training costs can be very high for reasoning models such as OpenAI GPT-o1 and o3 mini.

With the models running out of world data, and the model differentiation from the LLM providers becoming less and less differentiated, all of them are looking for ways to stand out. OpenAI has taken the direction of making their chatbot more human. OpenAI claims its EQ is better than other models, meaning the model recognizes tone and intent and responds with empathy rather than just solutions—just like a real human would. With that approach interactions can feel more natural and human like.

In initial testing, GPT-4.5  seems to show strong capabilities in agentic planning and execution, including multi-step coding workflows and complex task automation. However, there is concern in initial tests that it can fake emotional intelligence just like humans would.

However, at the current time these models are very expensive. GPT-4.5 is also restricted only to Pro users for now because of GPU shortage per OpenAI’s claim.”

• GPT 4.5 pricing: Price Input: $75.00 / 1M tokens Cached input: $37.50 / 1M tokens Output: $150.00 / 1M tokens
• GPT 4o pricing: Price Input: $2.50 / 1M tokens Cached input: $1.25 / 1M tokens Output: $10.00 / 1M tokens