As ChatGPT turns 3, Google and Anthropic steal the spotlight

ChatGPT hits three years

ChatGPT’s launch in late 2022 kicked off an AI race that reshaped tech and pushed massive investment into the space. Three years later, it looks far more capable than the simple chatbot people tried at first.

The newest versions now mix real-time information, better reasoning, stronger coding, and advanced image and video features.

That growth helped set the pace, even as the rest of the industry caught up. The question now is whether OpenAI can hold onto that lead with rivals closing the gap.

Google steals new momentum

Google was counted out early, but the picture looks very different now. Its latest Gemini updates impressed researchers and investors with strong performance across major benchmarks.

Its ability to analyze large codebases and handle agentic coding tasks showed how quickly Anthropic is maturing its technology.

This turnaround shifted attention back to Google, which benefits from a massive ecosystem that helps it scale AI features easily. Search, cloud, Android, and YouTube all support its AI distribution, creating an advantage that is tough for anyone else to match.

Anthropic rises even faster

Anthropic entered the race later but has made huge waves with its newest Claude models. Claude Opus 4.5 recently topped both GPT-5 and Google’s Gemini 3 on key agentic coding and coding-workflow benchmarks, surprising a lot of people watching the competition closely.

Its focus on reliability and safety also helped it stand out in the crowded AI market. The company also benefits from major partnerships, including a plan to use up to one million Google TPUs to scale its computing.

That gives Anthropic access to rare hardware at a time when compute shortages are slowing the entire industry.

Early excitement cools down

When ChatGPT first arrived, it dominated every conversation around AI. But as the field expanded, the tone grew more cautious. Investors are now paying closer attention to whether OpenAI can manage the cost of its massive AI infrastructure deals, which total over a trillion dollars.

Some researchers also note that key issues like hallucinations and reasoning gaps still appear, even in the latest models. Those limits raise questions about how far LLM technology can go and whether another approach will eventually take over.

Competition heats up quickly

The race no longer has a clear frontrunner. Google and Anthropic now compete directly with OpenAI on capability, speed, and product releases. Each company upgrades its models within months, making it harder for any one player to stay far ahead for long.

Experts say this rapid cycle is good for innovation because it forces every company to keep improving. It also leaves users with more choices and better features, which raises expectations across the board.

Gemini makes big comeback

At the start of the year, many saw Gemini as a lagging product. But Google completely reversed that view with strong updates and better performance across major evaluations. These gains helped boost investor confidence and renewed interest in Google’s AI plans.

Because Google controls so much of the tech stack, from search to chips, it can roll out improvements faster than many competitors. That advantage has made Gemini one of the most closely watched AI models right now.

Anthropic pushes coding power

Coding performance became a major battleground this year. Claude Opus 4.5 earned attention for topping its rivals on advanced coding benchmarks. Its ability to analyse large codebases and handle agentic coding tasks showed how quickly Anthropic is maturing its technology.

This shift also proves that the industry is no longer defined by one leader. Smaller but highly focused labs can take the top spot in specific areas, forcing everyone else to react quickly.

AI bubble fears grow

Rising concerns about an AI bubble have triggered selloffs across tech stocks. Investors are questioning whether the industry can sustain its current growth rate, especially with the huge costs of developing and running advanced models.

OpenAI is under more pressure than most because of its complex financing structure and long-term commitments. While new features like Sora and Pulse could bring more revenue, the market is watching closely to see if those opportunities turn into real returns.

Google’s chip edge grows

One reason Google moves so quickly is its custom TPU chips. These chips help train Gemini models and power huge services like search and YouTube. By using its own hardware, Google saves money and scales faster than companies relying on outside suppliers.

This hardware advantage also lets Google rent out computing power, creating an extra revenue stream. With AI models becoming larger every year, having this kind of control over compute is turning into a major competitive edge.

LLM limits spark debate

Researchers are questioning whether large language models can reach the next level of AI. Issues like hallucination and limited reasoning raise doubts about this approach.

Their reliance on old training data also concerns experts, and some believe new models will need a deeper understanding of information rather than simple prediction.

These concerns are shaping the conversation about what comes after today’s chatbots. They also affect investor expectations, since companies need clear paths to long-term automation for AI to deliver bigger returns.

World models gain attention

Some experts think world models could become the next big shift. These systems interact with physical spaces and build internal simulations of the world, which could help future robots, vehicles, and assistants understand their surroundings more naturally.

Supporters argue that this approach might be a better path toward AGI because it moves beyond text prediction. Although still early, world model research is picking up momentum as companies look for the next breakthrough.

ChatGPT still improves steadily

Even with new rivals, ChatGPT continues to evolve. Its ability to simulate understanding is far stronger than it was three years ago, making it more useful and easier for people to interact with. Researchers say these changes, while sometimes subtle, can be significant in real-world use.

As AI moves into everyday tools, even small improvements shape how people work, search, and create. Many expect ChatGPT and Gemini to keep upgrading in ways that feel incremental but have meaningful long-term impact.

Future fights move online

The next wave of competition may happen in AI-powered web browsers and search experiences. Companies like OpenAI and Perplexity are rolling out tools that reshape how users explore the internet. Google is also pushing more AI into search as it tries to retain its massive traffic.

Whoever wins this experience layer could control how people reach information online. That makes the next year especially important as new interfaces and assistants compete for attention.

To learn more about where AI infrastructure is heading, explore OpenAI’s plans for city-sized supercomputers.

Where the race goes next

Three years after ChatGPT arrived, the AI landscape is more crowded, more competitive, and moving a lot faster. Google and Anthropic now stand beside OpenAI rather than behind it, and new ideas like world models are shaping what comes after today’s chatbots.

The future will depend on who can deliver the best experience and scale responsibly. If you’re interested in how creators are responding to AI’s rapid evolution, you can read more about how some are using Sora to bring their characters to life.

What do you think this AI rivalry is heading toward? Share your thoughts in the comments.

Read More From This Brand:

• OpenAI’s ChatGPT might lose safety guardrails after long-term use
• Why OpenAI’s AI music generator could be a step too far?
• OpenAI identifies mental health concerns among ChatGPT users

Don’t forget to follow us for more exclusive content on MSN.

If you liked this story, you’ll LOVE our FREE emails. Join today and be the first to get stories like this one.

This slideshow was made with AI assistance and human editing.