Claude Opus 5.5 is the best AI for math. DeepSeek V4.1 Flash comes close for 11% of the cost, and Grok 4.7 answers fastest.
Claude Opus 5.5 is the best AI for math. DeepSeek V4.1 Flash comes close for 11% of the cost, and Grok 4.7 answers fastest.
20 models · Updated September 2026 · Data from Artificial Analysis
Claude Opus 5.5
Anthropic · High effort
The highest score of any model we compared at this setting.
DeepSeek V4.1 Flash
DeepSeek · Max effort
74% of Claude Opus 5.5's score for 11% of the cost.
Grok 4.7
SpaceXAI · Extra high effort
The quickest full answer from a model scoring 38 or more.
Claude Opus 5.5
Anthropic · High effort
The highest score of any model we compared at this setting.
DeepSeek V4.1 Flash
DeepSeek · Max effort
74% of Claude Opus 5.5's score for 11% of the cost.
Grok 4.7
SpaceXAI · Extra high effort
The quickest full answer from a model scoring 38 or more.
Each job favors a different model. Pick the one closest to yours.
Proofs and competition-style problems
Use
Claude Opus 5.5
Anthropic · High effort
The strongest reasoning at everyday effort. For the hardest problems, switch it to max effort: it scores 57.6 there, at a higher cost and longer wait.
Proofs and competition-style problems
Use Claude Opus 5.5
The strongest reasoning at everyday effort. For the hardest problems, switch it to max effort: it scores 57.6 there, at a higher cost and longer wait.
Homework help and practice problems
Use DeepSeek V4.1 Flash
$11.02 per 1,000 problems, about 11% of what Claude Opus 5.5 costs. It scores 39.5.
Quick calculations and checks
Use Grok 4.7
A full answer in about 12 seconds. Ask it to show its working so you can spot a slip.
Code that does the math
For heavy calculation, have a model write and run code, so you get a result you can check.
Each dot is a model. Higher scores better, further left is cheaper, so the best deals sit top left. The cost axis is logarithmic: every step left is a big saving. Hover a dot for its numbers.
Sort by the pillar you care about, and set your monthly volume to see the bill. Costs include each model's thinking tokens at the setting shown.
Claude Opus 5.5. At everyday effort it scores 53.6 against 51.2 for Claude Fable 5.1, and costs $98.51 per 1,000 tasks against $224.
Claude Sonnet 5 looks cheaper on list price ($2/$10 per million tokens against $4/$20), but at extra high effort it thinks so much that 1,000 tasks cost about $117, more than Claude Opus 5.5, for a score of 34.4.
Hard problems reward thinking longer. Claude Opus 5.5 goes from 53.6 to 57.6 at max effort. GPT-6 Sol goes from 44.1 to 47.5 at max effort, with answers taking about 171 seconds.
For a single hard problem, that wait and cost are usually worth it. For hundreds of routine problems, stay with everyday settings. Use the Max effort switch above to compare.
Every model on this page can reason step by step before answering. We compare every model at the setting people run day to day: its strongest reasoning setting that gives a full answer within 45 seconds in Artificial Analysis's tests. Switch to Max effort above for the benchmark numbers.
Intelligence is the Artificial Analysis Intelligence Index at that setting. It includes math and science reasoning alongside coding and knowledge, and it's measured the same way for every model.
Cost starts from a typical problem of 15,000 tokens, most of it the model's working. We scale the output by how much each model wrote, thinking included, in Artificial Analysis's tests, so a model that thinks a lot pays for it.
Speed is the median time to a full answer in the same tests.
Best value is the most intelligence per dollar and fastest is the quickest full answer, both among models scoring 38 or more (within 30% of the leader). Neither pick goes to a model its maker has replaced.
Claude Opus 5.5. It scores 53.6 on the Artificial Analysis Intelligence Index at everyday effort and 57.6 at max effort, the highest of the models we compared. DeepSeek V4.1 Flash is the value pick at $11.02 per 1,000 problems.
Claude Opus 5.5. At everyday effort it scores 53.6 against 51.2 for Claude Fable 5.1, and costs $98.51 per 1,000 tasks against $224. Claude Sonnet 5 looks cheaper on list price ($2/$10 per million tokens against $4/$20), but at extra high effort it thinks so much that 1,000 tasks cost about $117, more than Claude Opus 5.5, for a score of 34.4.
Claude Opus 5.5, especially at max effort. Thinking longer raises scores on hard reasoning, so a model at max effort can outscore a stronger model at everyday settings.
Grok 4.7, with a full answer in about 12 seconds. It scores 46.4, so check its working on anything that matters.
Answer three questions and get a pick for each part of your workload, with the monthly cost.