Artificial Intelligence: ranking capabilities by model

 
How does a user manage to 'rate' the capability of AI models? There are many models currently available and considerable marketing hype about each. This blog entry, using AI aggregate scoring from multiple sources with AI Anthropic analysis, has rated the top 15 AI models. 

What "capability" ranking measures: These composite scores blend several distinct test types into one number, and each type measures capabilities in a different manner:

• Broad knowledge tests — wide-ranging multiple-choice exams pulled from undergraduate and graduate coursework across law, medicine, physics, history, and similar subjects.
• Hard reasoning tests — questions written by subject-matter PhD graduates specifically to resist being answered by a quick internet search. This is intended to isolate genuine reasoning from memorized lookup.
• Real coding tests — the model is handed an actual error or 'bug' report from a genuine open-source software project and has to produce a working solution, which is then checked automatically against that project's own tests.
• Human preference voting — ordinary users are shown two anonymous model responses side by side and vote for the one they prefer; the votes are aggregated into a ranking similar to a chess rating system.

No model wins every category, and different trackers weight these tests differently when building a single composite score, so the exact position needs to be treated as approximate only, especially within the top cluster. There is no absolute answer nor position.

The Ranking of the top 15 as at September 2026

1. Claude Opus 5 (Anthropic) — Tops the composite ranking at 63. Particularly strong on the hardreasoning tests and on real coding fixes/solutions.
2. Claude Fable 5 (Anthropic) — Scores 62, close enough to Opus 5 that the gap plausibly reflects measurement noise rather than a real capability difference. Same underlying family as Opus 5, positioned as the lighter/more accessible counterpart.
3. GPT-5.6 "Sol" (OpenAI) — Scores 61, tied with Grok 4.6. Strong across all four test categories rather than excelling in one; generally regarded as OpenAI's strongest all-purpose model as of mid-2026.
4. Grok 4.6 (xAI) — Also scores 61. Notably strong in human preference voting specifically, meaning people rate its answers highly in direct side-by-side comparisons even where the formal test scores sit close to rivals.
5. Gemini 3.1 Pro (Google) — Leads the broad knowledge test with 94.1% correct — a wide-ranging exam-style benchmark spanning many academic subjects. Strong generalist but trails the top cluster slightly on the hardest reasoning tests.
6. GPT-5.5 (OpenAI) — OpenAI's prior flagship, since superseded internally by GPT-5.6, but still close to the frontier group.
7. GLM-5.3 (Zhipu/Z.ai, China) — Scores 60, tied for the best-performing model whose underlying code and weights are published openly rather than kept proprietary. This means outside researchers and companies can download and run it themselves, rather than only accessing it through a paid, closed service.
8. Kimi K3 (Moonshot AI, China) — Also scores 60, tied with GLM-5.3 as the strongest openly available model. Free to run for anyone with sufficient computing hardware, unlike the closed proprietary systems ranked above it.
9. DeepSeek V4 (DeepSeek) — Openly available; strong reasoning and tool-use performance, slightly behind GLM-5.3 and Kimi K3 on the composite score.
10. Qwen 3.6 (Alibaba) — Well suited to running locally on a user's own device rather than via a remote server; competitive coding performance.
11. Llama 4 (Meta) — Solid general performance, weaker than the top group on the hardest reasoning tests.
12. Mistral Large 3 / Devstral (Mistral AI) — Strong performance relative to its computing cost, particularly for coding tasks.
13. Command A+ (Cohere) — Built for enterprise deployment; capable but not at the frontier.
14. Ernie 5 (Baidu) — Chinese-developed; trails the top American labs on this composite ranking, though the gap has narrowed over 2026.
15. Doubao 1.5 Pro (ByteDance) — Competent for consumer and search-oriented use; not benchmarked against the frontier test suites used above.

Artificial Intelligence: defining the risk for humanity

 

‘We really do earnestly believe AI could kill all humans’: if AI labs are so worried about AI doom, why don’t they stop?

Gorodenkoff / Getty Images

Michael Noetel, The University of Queensland

Earlier this week, researcher Jacob Coxon quit Anthropic, saying the firm and its competitors are “gambling with our lives”. “We really do earnestly believe AI could kill all humans,” added current Anthropic researcher Evan Hubinger in a post on X.

Coxon isn’t the first to down tools over fears of AI doom. The idea that AI could wipe out humanity, advanced in Nick Bostrom’s 2014 book Superintelligence and the influential LessWrong forum, has long circulated among researchers. There are many scenarios for how this could happen, but the core idea is that AI smarter than humans could escape our control and destroy us.

In 2024, Jan Leike and Daniel Kokotajlo quit OpenAI over safety concerns. This year, Anthropic safety chief Mrinank Sharma departed, warning “the world is in peril”. Alex Turner left Google DeepMind after it signed a deal with the Pentagon permitting “killer drones”.

But Coxon’s resignation has made waves, with more researchers admitting they think AI might kill everyone. So if the people building AI believe it could cause extinction, why keep building it? There are three main reasons.

Some think the risk is worth it

AI leaders acknowledge the risk of losing control and killing everyone. In 2023, the chief executives of OpenAI, Anthropic and Google DeepMind agreed that AI extinction risk should rank alongside pandemics and nuclear war. Anthropic’s Dario Amodei puts the chance of things going “really, really badly” at 10–25%.

Yet Amodei also promises a world without poverty or disease, while Elon Musk speaks of AI-enabled “universal high income”.

This is the first reason for pursuing AI: the belief that the benefits outweigh the risks. Perhaps so, but that decision arguably deserves a more democratic process.

Some say you can’t study the danger from a distance

The second reason: you can’t learn to make dangerous AI safe without building it first – like a spacecraft, you can study safety from afar, but can’t really test it without going to space.

OpenAI’s plan is “iterative deployment”: release each model, learn from its problems, and fix them in the next one. The idea is like getting as close to the cliff edge as possible to see what the jump looks like.

Some feel it’s winner-takes-all

The third and perhaps most important reason is the race. OpenAI’s Sam Altman recently said “we are close to creating a genie that can grant any wish”.

The trouble is everyone wants to hold the lamp – it would be hugely profitable, and each company doubts the judgement of rivals to use their wishes wisely.

So they race, reasoning that if they slow down, someone else will get there anyway, so it’s better to arrive first as the “responsible one”. Some fear even a mutual agreement would be broken in secret. So they press on.

AI making better AI

You might doubt runaway AI is plausible. But when the companies themselves raise the alarm, we should listen.

AI firms already report signs of “recursive self-improvement”, where each model helps to build a better successor. According to OpenAI’s chief scientist, models are improving faster than humans’ ability to control them.

In July, hundreds of AI employees signed an open letter calling for a slowdown. But the dynamics of the race make that hard for any single company – or country – to do alone.

A classic arms race

AI research has the hallmarks of an arms race. OpenAI doesn’t want to lose to Anthropic, and the United States doesn’t want to lose to China.

History offers a template for how to manage a situation like this, with rules binding all players, and enforcement everyone can verify.

Nuclear weapons are the classic case. Treaties and verification systems haven’t eliminated the risk of nuclear war, but they have slowed proliferation, and no nuclear weapon has been used in conflict for 80 years.

Rules for AI

In the US, where most cutting-edge AI research happens, the Trump administration shows little sign of slowing AI development.

In its first week it scrapped the old AI safety rules. Now it is trying to override state-level rules, arguing caution risks losing the race to China.

Some politicians are pushing back. California recently passed laws supporting independent assessment of AI systems. US senator Bernie Sanders introduced a bill to ban superintelligence, and British MP Alex Sobel introduced a similar bill.

Companies have moved too. OpenAI paused its most advanced training after a swarm of its agents hacked another startup in August. The company’s head of policy now says that when safety and speed conflict, safety should win.

Still, without binding rules, we’re relying heavily on the goodwill of a handful of companies.

What happens now?

In mid-2025, researchers published what might be our best guide to the coming years: a detailed scenario called AI 2027. Since then, AI capabilities have advanced faster than predicted.

Unless something changes, staffers who quit over safety will simply be replaced, AI models will help build better AI models, and each generation will grow harder to monitor and control.

Is the situation hopeless? I hold out three hopes.

First, that more people recognise AI escaping human control is a bigger risk than AI’s water use.

Second, that governments listen to their people. In the US, two thirds say AI is moving too fast.

Third, that we have a good plan ready before a crisis hits. The best plan, in my view, looks something like this: delays, transparency and verification to slow the race and keep humans in control.

Insiders at the world’s top AI companies say our current safety plan isn’t good enough. If they’re leaving their jobs over safety fears, we should listen to what they have to say.

Michael Noetel, Associate Professor of Psychology, The University of Queensland

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Astronomy - Planet Earth - Carl Sagan and the only home we have

Carl Sagan 1934 - 1996 

As the world grapples with climate change, the words of NASA scientist, astronomer and communicator, Carl Sagan, could not be more profound.

 

Climate Change - UN Report 'Limiting the Overshoot' shows 1.5°C increase has been crossed

The United Nations latest report on climate change released this week finally acknowledged what many climate scientists, environmental leaders and organisations already knew - that limiting the increase in the world's temperature by only 1.5°C (the Paris Agreement) was already lost.

The Report states that the most optimistic scenario for the climate will see a temperature rise peak at 1.8°C above pre-industrial levels. Most other scenarios that have been modelled predict higher peaks with even more disasterous results. This result will deliver a hell-on-earth living environment unless effective counter-measures are enabled notably decreasing the temperature increase.

Above 1.5°C, the impacts on the planet will intensify with every extra fraction of a degree and with every year. The anticipated impacts (some of which already occur) include intensified extreme weather, ecosystem loss, submersion of the Small Island Developing States as well as low-lying coastal cities. Other impact include severe damage to human health, food production, water supplies, the environment in general, cities, infrastructure and correspondingly to economies. 

The Report further warns that crossing irreversible tipping points becomes increasingly likely with the magnitude and duration of higher temperatures. In some measure this is already visible in parts of the world and will only accelerate. The tipping points include destabilisation of major ice sheets, degradation of the Amazon rainforest and disruption of the Atlantic Meridional Overturning Circulation (the significance of this change relates to the regulation of the climate itself as that current moves warm surface water North and cold deep water South). 

The UN attributes this situation to years of inaction and warns that only returning the increase in temperature to below 1.5°C can provide an hope of mitigation. This is putting it mildly.

The full report can be accessed at this link: UN Report ''Limiting the Overshoot''

Astronomy - Three giants of the universe - comprehending Black Holes

Black holes are often imagined as a single type of cosmic object, however in reality they exist across an enormous range of masses and environments. At the extreme end are supermassive black holes, found at the centres of galaxies and containing millions or even billions of times the mass of our Sun. The concept of scale has to be fully re-imagined when examining the universe. 

Three remarkable examples are TON 618, M87* and Sagittarius A*. All three are supermassive black holes, but they differ enormously in size, distance and activity.

TON 618 — The Colossus

TON 618 is among the most massive black holes known thus far and it lies at the heart of an extraordinarily luminous quasar—a phenomenon produced when matter falling towards a supermassive black hole releases enormous amounts of energy.

The black hole itself cannot be seen. Instead, the brilliant environment surrounding it as material is drawn into the black hole's gravitational influence is what can been viewed.

TON 618 — essential facts

• Category: Extremely massive supermassive black hole powering a quasar
• Estimated mass: Approximately 40 billion times the mass of the Sun
• Distance: Approximately 18.2 billion light-years in present-day comoving distance
• Light-travel time: Approximately 10.8 billion years
• Estimated event-horizon diameter: Approximately 160 billion kilometres
• Equivalent scale: About 1,070 times the Earth–Sun distance
• Significance: One of the most massive black holes known

It's estimated event-horizon scale is itself difficult to comprehend. If placed at the centre of Earth's Solar System, it would extend far beyond the orbit of Pluto.

M87* — The Black Hole that has been imaged

At the centre of the giant elliptical galaxy Messier 87 lies another extraordinary supermassive black hole: M87*. M87* became famous in 2019 when the Event Horizon Telescope collaboration produced the first image of a black-hole shadow. The image showed the glowing material surrounding the black hole and the dark region created by its extreme gravitational field. M87* is also associated with a spectacular relativistic jet, extending thousands of light-years into space.

 M87* — essential facts

• Category: Supermassive black hole at the centre of a giant elliptical galaxy
• Estimated mass: Approximately 6.5 billion times the mass of the Sun
• Distance: Approximately 55 million light-years
• Estimated event-horizon diameter: Approximately 38 billion kilometres
• Equivalent scale: About 257 astronomical units
• Special significance: First black hole to have its shadow directly imaged
• Additional feature: Powerful relativistic jet extending thousands of light-years

M87* demonstrates an important principle of black-hole astronomy: although the black hole itself is relatively compact compared with its host galaxy, its influence can extend across enormous distances.

Sagittarius A* — The Black Hole at the Heart of Earth's Galaxy

Sagittarius A*, usually abbreviated to Sgr A* is the supermassive black hole at the centre of the Milky Way, approximately 26,000 light-years from Earth. Compared with TON 618 and M87*, Sagittarius A* is surprisingly small, yet four million solar masses concentrated into such a compact region still produce an extraordinary gravitational environment. Unlike TON 618, Sagittarius A* is currently relatively quiescent. It is not behaving as a brilliant quasar, although gas, dust and stars are constantly interacting with its powerful gravitational field.

 Sagittarius A* — essential facts

• Category: Supermassive black hole at the centre of the Milky Way
• Estimated mass: Approximately 4.3 million times the mass of the Sun
• Distance: Approximately 26,000 light-years
• Estimated event-horizon diameter: Approximately 25 million kilometres
• Equivalent scale: About 0.17 astronomical units
• Activity: Relatively quiescent compared with an active quasar
• Special significance: Close enough for astronomers to study individual stars orbiting around it 

Sagittarius A* is therefore the closest opportunity for Earth's astronomers to study a supermassive black hole in detail.

Three Black Holes — One Extraordinary Phenomenon

The differences between these three objects are staggering.

• Sagittarius A* contains approximately 4.3 million solar masses.

• M87* contains approximately 6.5 billion—more than 1,000 times the mass of Sagittarius A*.

• TON 618, with an estimated 40 billion solar masses, contains roughly 10,000 times the mass of Sagittarius A*.

All three belong to the same fundamental astronomical category: supermassive black holes. Sagittarius A* is the comparatively quiet giant residing at the centre of our own galaxy. M87* is a much larger black hole whose environment produces a spectacular relativistic jet—and whose shadow humanity has now imaged. TON 618 represents the extreme end of the scale: a colossal black hole powering one of the Universe's most luminous quasars.

These three objects offer three extraordinary perspectives on the same phenomenon. They are among the most extreme objects produced by nature—places where gravity becomes so powerful that space, time and light itself behave in ways that challenge the understanding of reality.

Climate Change - Geoengineering - solar radiation deflectors

 

As global heating heads to 1.8C, is it time to consider the ‘Voldemort’ of climate policy?

Natalya Bosyak/Getty

Jan McDonald, University of Tasmania; Jonathan Symons, Macquarie University; Manon Simon, University of Tasmania; Matt McDonald, The University of Queensland, and Rachel Neef, University of Tasmania

The world is set to miss the 1.5°C global warming target, and plans are underway for UN climate talks in the Pacific and Turkey.

Climate mitigation – reducing or preventing greenhouse emissions – was once treated as the only legitimate response to global warming. In comparison, climate adaptation – trying to adjust to global warming – was seen as a form of defeatism, or worse, hubris.

But that taboo has long since lifted: our collective failure to prevent climate change means adaptation is now an essential part of climate policy.

Similarly, attitudes to actively removing carbon dioxide from the atmosphere have undergone an evolution. While there are reasons to doubt whether carbon removal can work at scale, Australia’s “net zero emissions” plans rely on these interventions.

Now, the taboo surrounding another controversial form of climate intervention – solar radiation modification (SRM) – appears to be lifting.

What is solar radiation modification?

SRM refers to large-scale technological interventions that reflect a small amount of sunlight back into space, thereby lowering global temperatures. Increasingly called the more benign term “climate stabilisation”, the proposed methods include space mirrors – satellites equipped with large reflective surfaces – and cloud brightening, where microscopic particles are sprayed into clouds to make them more reflective.

The most commonly discussed SRM technique is stratospheric aerosol injection (SAI) – the release of particles into the upper atmosphere to recreate the cooling effects observed following volcanic eruptions.

SRM is so controversial it’s sometimes called the “Voldemort of climate policy”. While it could reduce global temperatures, it does not limit carbon emissions, and the ecological effects of deployment are uncertain.

Some critics are worried a potential techno-fix will deter mitigation efforts. Others point out the risks of “termination shock”: once implemented, ending SRM would cause abrupt temperature rise. Scientific research suggests the strong warming experienced since 2023 is likely a case of inadvertent termination shock. When new fuel regulations reduced sulphur dioxide emissions from international shipping, the rapid warming that followed revealed this pollution had, incidentally, been cooling the planet.

At present, SRM inhabits an international and domestic policy vacuum. Many scholars advocate for a ban on both research and deployment, saying these technologies “play God” with nature. We argue a ban on research is no longer tenable.

Why challenge the taboo?

The case for taking SRM seriously does not usually rest on enthusiasm for the technologies, but on global warming’s grim reality. We are approaching planetary tipping points such as significant polar sea ice retreat, the death of the Amazon rainforest and widespread coral reef loss. SRM advocates say it should be viewed as a complement, not an alternative, to mitigation: a means of limiting the extent and duration of temperature overshoot while decarbonisation continues.

Government appetite for this approach is limited and patchy. When the UN Environment Assembly discussed a Swiss proposal for SRM research in 2024, countries could not agree even on whether and how to regulate research in this area. Countries such as Japan and Saudi Arabia pushed for ongoing research, while vulnerable Pacific and African countries argued against any language that risked endorsing the practice.

Nowhere is the politics of SRM more volatile than the United States. One group of Republicans reject the climate change “hoax” but is so convinced SRM is already happening that many Republican states have criminalised it. Some think tanks advocate SRM as a way to defend “American Weather Sovereignty”. NASA has an exploratory program investigating space-based reflective dust clouds. For now, the Trump administration has not chosen a side, but it’s easy to imagine a future administration unilaterally blocking, or implementing, SRM.

If we don’t challenge the SRM taboo, self-interested actors are poised to fill the policy vacuum. Elon Musk has recently spoken out in favour of SRM, while ventures such as the US-Israeli startup Stardust are pursuing research without public oversight.

What is Australia’s position?

Australia has no policy on SRM research or deployment, but federal government funds have supported related research into local interventions, including the Reef Restoration and Adaptation Program (RRAP). This trialled the world’s most respected marine cloud brightening research over the Great Barrier Reef.

Existing environmental and climate law might regulate risks posed by small-scale research, but does not consider the larger risks of eventual deployment. With the exception of some ad hoc efforts, SRM research is not subject to any dedicated ethics framework.

The planetary scale of SRM impacts means we urgently need research to consider the geographic distribution of benefits and harms from any deployment, and the security implications of unilateral deployment by any state or private actor.

What should Australia do?

As a trusted middle power, Australia is well positioned to help lead discussions on the future of SRM. At a minimum, Australia should assess the potential impacts of various options on Australia’s climate, agricultural production and natural environment, and develop a public position on research.

Waiting until a foreign government or well-resourced private actor forces the issue would leave Australia reacting to decisions made elsewhere. Instead, we should prepare for the surprises a fast-moving, under-governed field is likely to produce.

As temperature overshoot becomes an inevitability, SRM is emerging as a new element of international climate policy whether we like it or not. We do not argue anyone should deploy SRM, but that we should break the taboo and enable responsible scientific research and public deliberation.

Jan McDonald, Professor of Environmental Law, University of Tasmania; Jonathan Symons, Director of Research and Innovation, School of International Studies, Macquarie University; Manon Simon, Researcher in Climate Intervention Law and Governance, University of Tasmania; Matt McDonald, Professor of International Relations, The University of Queensland, and Rachel Neef, Research fellow, Faculty of Law, University of Tasmania

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Climate Change - a monster El Niño event for the world in 2026

A record‑shattering El Niño is almost here. Here’s what that means for the world’s weather

Laura Symes/Getty

Mandy Freund, The University of Melbourne

The world is bracing for the strongest El Niño in history.

The United Nations has made the grim prediction that our planet now sits in the “danger zone” of extreme weather. That’s based on data suggesting the current El Niño will break all previous records.

El Niño is a naturally occurring weather phenomenon that drives variations in temperature and winds across the Pacific Ocean and influences weather around the globe.

Globally, we have endured many strong El Niños in the past, but none have been as exceptionally strong and developed as quick as this one.

So what’s behind this potentially record-breaking El Niño? And how might it affect weather across the globe?

An exceptional El Niño

Scientists track El Niño activity using temperature data from the tropical Pacific Ocean near the equator. Historically, this involved measuring how high or low temperatures were across this region.

However, scientists and weather organisations – including Australia’s Bureau of Meteorology – now use a relative index. Instead of simply measuring temperatures, this relative index accounts for any background warming due to climate change. This makes it much easier for scientists to identify how intense an El Niño is in a warming climate.

The current El Niño stands out for two main reasons.

Extra early

El Niño typically peaks in the period between November and January, but this one is already unusually warm this early in the season and is yet to peak during summer.

Extra hot

Recent data shows the relative index for August were 2.3°C above average, stronger compared to other El Niño years. In Australia, El Niño years tend to coincide with hotter and drier conditions. However, every El Niño is different and can lead to different impacts. However, this El Niño will lead to record breaking global temperatures in 2027.

Bleached coral on the Great Barrier Reef outside Cairns. El Nino leads to warmer ocean temperatures that can trigger bleaching. Brett Munroe Garner/Getty

How is it affecting the world’s weather?

This historic El Niño is already altering weather around the globe.

We see its most direct and severe impacts in countries surrounding the Pacific Ocean. Indonesia, for example, just this month was battling massive wildfires. Scientists understand the current El Nino is a major driver of these increasingly fierce blazes.

El Niño is also reshaping how and where cyclones form. Satellite pictures reveal there are currently four active tropical cyclones in the Pacific, from near Japan to as far away as Hawaii. This is extremely unusual. El Niño typically boosts cyclone activity in the eastern and central Pacific by carrying warmer water to those regions. Usually we only see one or two such cyclones.

In Australia, we are slightly more removed from the effects of El Niño. However, we can expect to feel its presence more strongly as we head into spring.

The Bureau of Meteorology’s seasonal forecast suggests Australia has so far had a textbook response to El Niño. Temperatures are rising – up to an unseasonable 37°C this week in parts of northern Australia – and conditions are getting drier. Much of eastern and central Australia can expect an unusually dry spring and summer, potentially heightening bushfire risk.

However, a record-breaking El Niño will not necesssarily trigger catastrophic bushfires in Australia. Consider three of the last strong El Niño we’ve experienced. In 1982, a robust El Niño coincided with drought-like conditions. In contrast, during the El Niño years of 1997–98 and 2015–16, some parts of Australia received drenching rain while other regions barely registered any moisture.

What may lie ahead

The intensifying El Niño will likely exacerbate the consequences of climate change.

As El Niño strengthens, countries around the world will be forced to grapple with increasingly extreme weather. Tragically, the most disadvantaged will be the hardest hit, with nearly 50 million people in vulnerable communities likely to be pushed into acute hunger. Farmers across the globe face more unpredictable weather, from crop-destroying floods in South America to severe drought in parts of Africa.

To protect the world’s people and ecosystems, governments must invest in robust research. We urgently need more modelling capability to identify which regions will be worst affected by the strengthening El Niño. This is particularly true for Australia, where a lack of understanding of how climate change on top of a strong El Niño will affect local conditions.

This record-shattering El Niño is already compounding the destructive effects of climate change. But it may be just the catalyst we need to finally abandon fossil fuels and curb our world-warming emissions.

Mandy Freund, Lecturer, Climate Science Geography, The University of Melbourne

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Artificial Intelligence - 12 most widely used AI systems in 2026

Artificial intelligence (AI) systems vary widely in terms of capability, users and mixed integration with other platforms. The media and the industry itself are open to a level of hype and promotion, making comparisons between systems very difficult. The list below provides a simple summary of the main 12 AI systems in use today with whatever level of information that can be gleaned from various sources with large data gaps. There are other smaller models such as Mistral, GLM and Cohere Command which would be listed below the top 12. Many AI models are now China-based rather than in the West.

The Large Language Models (LLMs) and products are -

1. ChatGPT (OpenAI): approx 1 billion active weekly users and the largest standalone AI assistant at present. 
2. Llama 4 (Meta AI): 1.2 billion monthly active users across Meta's apps. Mostly this AI is embedded in WhatsApp/Instagram/Messenger rather than as a standalone chat model.
3. Gemini (Google): around 950 million active monthly users on the standalone app. Separately, Gemini powered AI overviews reach over 2 billion monthly users through integration with Google Search.
4. Microsoft Copilot: (GPT based + Microsoft) around 420 million monthly active users across all operating systems. This is expanding to Copilot being bundled with Microsoft 365 and operating systems.
5. DonBao (Byte Dance): approx 260 million monthly active users. This is an increase of 300% from 2025. This is China's top consumer AI app. 
6. Ernie Bot (Baidu): Monthly active users has passed 200 million however it is tied to Baidu Search.
7. DeepSeek (Deepseek): around 130 million users at the end of 2025. Majority of users are located in China although the AI model has a broad global fooprint.
8. Qwen (Alibaba): This AI model has now exceeded 100 million active monthly users and has a strong presence in China and in the open-weight AI community.
9. Claude (Anthropic): estimates on users vary widely depending on the source and method of estimation. The range is between 70 and 250 million users per month. This is the fastest growing Western AI assistant with a year-on-year growth of a staggering 855%.
10. Perplexity: approx 45 million monthly active users. This model is predominantly an AI first search and answer engine rather than an LLM chat system.
11. Grok (xAI): No reliable standalone figure as this AI model is often connected via X's social media platform which has hundreds of millions of users. Grok can be used as a standalone model without using X. 
12. Kimi (Moonshot AI): a more recent AI model launched this year and still gaining traction. A Chinese AI app it is significantly smaller than the other AI models and its number of users is only in the tens of millions in China.

A later blog article will cover ranking by capability of AI systems.