Skip to content

    OpenAI

    GPT-6.1 Sol

    Released

    GPT-6.1 Sol is the newest Sol model from OpenAI, released in September 2026. It ranks number 3 of 79 with a score of 9.6 out of 10, among the very best available. Its best areas are coding (9.6/10) and business and finance (9.0/10); creative writing (7.9/10) is the weakest of the nine tasks we track. In blind votes against Gemini 4 Argon, the model people prefer most, it was picked about 44 times in 100. It needs a paid plan, from $20 a month. For developers it costs more than about 75% of the models we rank.

    9.6/10Best available
    Compare

    Rank 3 of 79

    What it's good for

    Coding9.6

    Writing, fixing and explaining code

    Business and finance9.0

    Plans, analysis, spreadsheets and money questions

    Science and research9.0

    Explaining and researching science and social topics

    Maths8.9

    Calculations, puzzles and step-by-step problems

    Health questions8.8

    General health information (not medical advice)

    Following detailed instructions8.7

    Doing exactly what you ask, in the format you ask

    Other languages8.7

    Working in languages other than English

    Writing and editing8.4

    Emails, reports, rewriting and proofreading

    Creative writing7.9

    Stories, poems, speeches and ideas

    Out of 10, compared with the best model at each task

    OpenAI

    GPT-6 Astra

    9.5/10 Best available

    GPT-6 Astra is the newest Astra model from OpenAI, released in September 2026. It ranks number 5 of 79 with a score of 9.5 out of 10, a very strong result. Its best areas are coding (9.8/10) and business and finance (8.9/10); creative writing (8.0/10) is the weakest of the nine tasks we track. In blind votes against Gemini 4 Argon, the model people prefer most, it was picked about 43 times in 100. It needs a paid plan, from $20 a month. For developers it is among the priciest models we rank.

    From $20 a month

    OpenAI

    GPT-5.6 Sol

    8.5/10 Very strong

    GPT-5.6 Sol is an earlier OpenAI model from July 2026; the newer GPT-6.1 Sol is also in our ranking. It ranks number 10 of 79 with a score of 8.5 out of 10, a solid result. Its best areas are coding (9.4/10) and business and finance (9.2/10); creative writing (8.5/10) is the weakest of the nine tasks we track. In blind votes against Gemini 4 Argon, the model people prefer most, it was picked about 45 times in 100. It needs a paid plan, from $20 a month. For developers it costs more than about 88% of the models we rank.

    From $20 a month

    OpenAI

    GPT-6 Sol

    8.3/10 Solid

    GPT-6 Sol is an earlier OpenAI model from September 2026; the newer GPT-6.1 Sol is also in our ranking. It ranks number 11 of 79 with a score of 8.3 out of 10, a solid result. Its best areas are coding (9.1/10) and health questions (8.2/10); maths (7.3/10) is the weakest of the nine tasks we track. In blind votes against Gemini 4 Argon, the model people prefer most, it was picked about 41 times in 100. It needs a paid plan, from $20 a month. For developers it costs more than about 75% of the models we rank.

    From $20 a month