Which AI model to use? (Sept 2026)

Every AI lab says its newest model is the best one. Most of them are right about something. This page sets out, category by category, which of the September 2026 models actually leads on medical, coding, legal, cybersecurity, reasoning, images, video and more — with the published scores, the price, and the caveats the launch posts leave out.

The short answer

No single model wins everything. The right answer depends on the job.

The models compared

These are the current generally available flagships from each lab, plus the leading open-weight models. Claude Mythos 5.1 and Grok 4.7 are deliberately absent: Mythos is restricted to vetted US organisations, and Grok 4.7 had not shipped at the time of writing despite several announced dates.

Category rankings

Reasoning and overall intelligence

The Artificial Analysis Intelligence Index v4.3 combines ten tests, including agentic knowledge work, terminal tasks, document reasoning and factual accuracy, with 40% of the questions held private so vendors cannot train on them. The current ceiling is 53. Bars here are scaled to the leader.

If you have seen index scores of 57, 61 or 66 quoted for these models, those are older scales from earlier in September. Only v4.3 numbers are comparable with each other.

Science knowledge

GPQA Diamond is a set of graduate-level physics, chemistry and biology questions written so that experts outside the field cannot answer them with Google. The frontier is now bunched within six points of the ceiling.

Maths

FrontierMath Tier 4 is research-level mathematics. Note that these are OpenAI-reported figures and that OpenAI funded FrontierMath and had sight of many of its problems, so treat the gap as directional rather than settled.

Coding

Terminal-Bench 2.1 is the widest independent comparison: agentic tasks run in a real shell, scored by Artificial Analysis on the same harness for every model. It is the fairest single number for "can this thing do engineering work".

The harder tests tell a different story

DeepSWE 1.1 puts a model inside a real repository and asks it to fix real issues. Terminal-Bench 4.0 is the newer, much harder version of the shell test. On these the order changes and the field spreads out.

What about PHP, C# and SQL?

Every headline coding score above is Python-biased: SWE-bench Verified is 500 tasks drawn from twelve open-source Python repositories, and DeepSWE and Terminal-Bench lean the same way. Nobody publishes a PHP or C# leaderboard. Two datasets get close, and both point the same way.

Opus 4.6 dropped only three points moving from Python-only Verified (80.8%) to Multilingual (77.8%), and on the private commercial set Claude degrades less than its competitors when moving from public to unseen repositories. Both figures are from the previous Claude generation, because the current one has not been re-run on these tests; the pattern is the point.

Desktop and computer use

OSWorld 2.0 gives a model a real desktop and 108 long workflows that take a human around 1.6 hours each. The headline numbers are partial-credit; the strict "finished the whole job" rate is far lower and shown alongside.

Strict end-to-end completion: Fable 5.1 41.7%, Opus 5 39.6%. Fable scores zero on any task where its safety classifiers intervene, which drags its average down on security-adjacent work. Astra completes tasks around 47% faster than GPT-5.6 Sol, so it often wins on wall-clock time despite the lower score.

Global knowledge

Humanity's Last Exam is 2,500 expert-written questions across every academic field. With tools enabled, Claude leads and this is the one academic test GPT-6 Astra loses.

Gemini's figure is on the HLE-Verified variant, so treat it as approximate. Factual accuracy specifically is covered in the hallucination section below, where Fable 5.1 also leads at 67%.

Biology and life sciences

This is the thinnest category. Only OpenAI publishes head-to-head life-science scores, Anthropic routes biology questions on Fable to Opus 5, and the strongest biology model on either side (Claude Mythos 5.1) is not available in the UK.

Legal (professional)

Legal benchmarks are the most fragmented, and each one is led by a different model. The consistent finding across all of them: models pass around 90% of the individual criteria on a legal task but complete very few whole tasks end to end.

Reasonable working rule: Claude for drafting and precision, GPT for research, and a human reads everything before it goes out. None of these models should be trusted to complete legal work unsupervised.

Cybersecurity

Two tests matter. ExploitBench measures whether a model can turn a known vulnerability into a working exploit; CyberGym measures whether it can find vulnerabilities in real codebases. A model can be excellent at one and mediocre at the other.

ExploitBench above. On CyberGym (vulnerability discovery) the order flips: DeepSeek V4.1 Flash 88.1%, Gemini 3.8 Flash cyber variant 86.2%, GLM-5.3 84.5%. Claude Fable 5.1 has no published cyber score because penetration-testing, exploit and binary-scanning requests are routed to Opus 4.8; Gemini's cyber variant is only available inside Google's Fairwind programme; and Astra's public build declines advanced offensive work unless your organisation is enrolled in OpenAI's Daybreak scheme.

Tool use and agents

AutomationBench-AA is Artificial Analysis's version of Zapier's business-workflow test: multi-step tasks across real apps. The headline score gives partial credit; the strict figure counts only workflows finished without breaking a single rule.

Strict, no-rule-broken completion: Astra 41.6%, Fable 5.1 32.1%, Opus 5 28.3%. On AA-Briefcase, the multi-week knowledge-work test, the order reverses and Fable 5.1 and Opus 5 lead ahead of Astra and Muse Spark 1.3. Grok 4.6 is the value pick here: near the top at a third of the price.

Medical and health

Three different questions get asked under "medical", and they have three different answers.

Professional use: answering to a physician's standard

HealthBench Professional is built from 5,000 conversations and physician-written rubrics from 262 doctors in 60 countries, scored on the length-adjusted version so that longer answers do not automatically win. It is OpenAI's benchmark, and all figures are OpenAI-run.

Fable 5.1 scoring below Fable 5 is a safety-routing artefact, not a capability regression: questions its classifiers flag are answered by a smaller model and marked down accordingly. No Gemini, Grok or open-weight model has a published HealthBench Professional score.

General medical information for the public

No consumer-grade health benchmark has been published for this generation of models, so the professional ranking above is the best available guide, read together with the hallucination table further down. The practical change this month is on the refusal side: Anthropic reports that Fable 5.1's biology safeguards fire 85% less often on benign elementary biology and medical questions than Fable 5's did, which means fewer "I can't help with that" answers to ordinary health queries.

Scan and imaging interpretation

The imaging benchmarks exist, but they live in academic papers rather than on vendor leaderboards: Radiology's Last Exam (RadLE) tests chatbots against board-certified radiologists on hard spot-diagnosis cases, ReXVQA covers chest X-ray question answering, CXR-LT is a multi-centre chest X-ray challenge with over 145,000 images, and newer sets such as NeuroQA cover 3D brain MRI. No lab publishes scores on any of them, so the only figures come from independent studies — which run a generation behind the current models.

Other studies agree on the ordering. On pneumothorax from chest radiographs, ChatGPT-4o was most accurate at 69.6%, then Claude 3.5 at 64.9% and Gemini 2.0 at 57.4% — but every model fell to between 12% and 21% on children, and ChatGPT dropped from 81.6% on large pneumothoraces to 42.2% on small ones. On measuring liver metastases against a radiologist's reference, Gemini reached an agreement score of 0.81, GPT-o3 0.52 and Claude 4 Opus 0.07.

Claude's pattern is consistent: it is the weakest of the majors at reading pixels and the strongest at reading radiology text. On structuring 3,949 head CT reports, Claude was significantly more accurate than GPT and Gemini for intracranial haemorrhage, and on 56 JAMA neuroradiology cases Claude 3.5 achieved the highest accuracy (80.4%) when given the image and the report together.

Two caveats. Every study above tested general models one generation behind the current ones — GPT-6 Astra, Claude Fable 5.1 and Gemini 3.8 Flash have not yet been run through RadLE. And the RadLE cases are deliberately hard: on routine chest X-rays the chatbots do markedly better, as the pneumothorax study shows, and MedGemma does better still.

Speed

Two different things get called speed. Tokens per second is how fast text appears on screen; time per task is how long an agent takes to finish a job, which depends as much on how many tokens a model needs as on how fast it produces them.

If you are running agents, token efficiency is the number that sets your bill. Astra is slow per token but so concise that its cost and time per task frequently undercut faster-looking rivals.

Hallucination and accuracy

AA-Omniscience asks 6,000 hard factual questions and tracks two things separately: how often the model is right, and how often it makes something up when it does not know. A model can post a low hallucination rate simply by refusing to answer, so read the two columns together.

The very lowest hallucination rates on the board belong to small models that refuse most questions (Cohere's Command A+ at 14.2%, with 9% accuracy). That is not a model you want; it is a model that has learned to say "I don't know". Also worth knowing: reasoning modes hallucinate two to three times more than plain modes on summarisation tests. If the facts matter, turn thinking off or give the model a document to work from.

Refusals and over-blocking

This is the category businesses ask about most and the one with the least data. No leaderboard measures "refused a perfectly reasonable request" across vendors, so the table below collects what has actually been published.

A model that refuses to look at malware is no use to the people whose job is looking at malware.

Image generation

Image models are ranked by blind human votes in the Artificial Analysis arena: two images from the same prompt, pick the better one, thousands of times. Elo is the native unit; higher is better and 30 points is a meaningful gap.

For diagrams, technical illustrations and anything with words in it, no benchmark measures "diagram quality" directly; text-rendering ability is the proxy, and GPT Image 2/2.5, Ideogram 4.0 and Nano Banana Pro lead it. For architecture diagrams you will usually get a better result asking a language model for SVG or Mermaid than asking an image model to draw one.

Video generation

Chinese labs currently dominate video. Google's Veo 3.1 is the strongest option that is straightforward to license in the UK, and it is the only model producing properly synchronised 48kHz speech rather than just sound effects.

What it costs to use them yourself

API prices are in the model table above. For individual chat subscriptions, the whole market has settled on a $20-a-month standard tier, with budget tiers below it and power-user tiers of $100 to $300 above it. Prices are the vendors' US list prices; UK checkouts are billed in sterling with VAT added, so check the vendor page for the exact figure on the day.

Two things to check before relying on a consumer plan for work: some Gemini app features have excluded the UK and EEA at launch, and Claude Mythos 5.1 is not available in the UK on any plan. Business and Enterprise tiers from every vendor add admin controls, data-processing terms and, in most cases, EU or UK data residency.

What this means for a UK business

• Day-to-day development: Claude Opus 5 is the best capability per pound for PHP, C#, Python and SQL work. Keep Fable 5.1 or GPT-6 Astra for the hardest long-running jobs, and Gemini 3.8 Flash for bulk, low-stakes tasks.
• Security work: cloud models over-block. Use Opus 5 where the cloud is acceptable, and a self-hosted DeepSeek V4.1 Flash or GLM-5.3 for malware triage and vulnerability hunting, which also keeps client code inside the UK.
• Client data: prefer Business or Enterprise tiers with UK/EU data-processing terms, or open weights on your own hardware. A $20 consumer plan is not a data-processing agreement.
• Anything that leaves the building: legal, medical and client-facing text gets a human read every time, whichever model wrote it. Fable 5.1 is the most accurate model measured and it is still wrong about a third of the time on hard facts.
• Marketing assets: GPT Image 2.5 or Ideogram 4.0 for graphics with words in them; Veo 3.1 for video you can license in the UK; Kling 3.0 if cost per second matters more than provenance.
• Expect this page to change: Grok 4.7 is due imminently, Google's next Pro model is overdue, and the index has been re-based three times this month. We will update the figures as they move.