Best AI for coding 2026.

Today's AI models, ranked by how well they code on two tests from an independent evaluator. One with real changes to open-⁠source projects and one with complex work in the terminal.

Data from 13 November 2026.

  • The best at FrontierCode

    Claude Fable 5.2

    48.8% on FrontierCode

  • The best at Terminal-⁠Bench

    Kimi K3.1 Pro

    67.2% on Terminal-⁠Bench

  • The best open-⁠weights model

    DeepSeek V4.1 Flash 1017

    28.5% on FrontierCode

  • The cheapest per task

    Claude Haiku 5.5

    $0.41 per task, among the top ten

Sort by

58 models

#Model
1Claude Fable 5.2 Anthropic · max effort48.8%59.6%65$26.54$8 and $409 Nov 2026
2Claude Opus 5.5 Anthropic · max effort46.4%59.6%59$10.04$4 and $2022 Sept 2026
3GPT-6 Astra OpenAI · max effort42.4%59.1%57$6.24$10 and $503 Sept 2026
4GPT-6.1 Terra OpenAI · max effort42.3%55.6%51$0.96$1 and $526 Oct 2026
5Claude Sonnet 5.5 Anthropic · max effort41.7%63.6%55$10.02$2 and $1028 Sept 2026
6GPT-6.1 Sol OpenAI · max effort40.6%56.1%55$1.27$2 and $1029 Sept 2026
7Gemini 4 Carbon Google · high effort40.6%65.2%60$5.48$2 and $10 $4 and $20 from 1 December10 Nov 2026
8Gemini 4 Hydrogen Google · extra high effort39.0%53.5%54$3.16$1.50 and $7.5012 Nov 2026
9Muse Wonder Full Meta · max effort37.8%57.1%50No dataNo per-token price12 Oct 2026
10Claude Haiku 5.5 Anthropic · max effort34.9%32.8%36$0.41$0.10 and $0.507 Oct 2026
11Tencent Hy4 Tencent31.4%37.4%35$1.03$0.834 and $2.501 half price until 31 December4 Nov 2026
12Grok 4.7 xAI · extra high effort28.7%25.8%35$7.77$2 and $621 Sept 2026
13DeepSeek V4.1 Flash 1017 DeepSeek · max effort · open weights28.5%33.8%33$0.60$0.30 and $1.2017 Oct 2026
14Muse Wonder Meta · max effort27.9%52.0%44$6.95$4 and $2012 Oct 2026
15Muse Ember Meta · max effort27.2%40.9%36$0.14$0.10 and $0.509 Nov 2026
16Qwen3.8-Flash-Next Alibaba · open weights27.1%25.3%29$0.57No per-token price26 Aug 2026
17Gemini 4 Argon Google · high effort25.7%57.1%50$4.97$2 and $10 $4 and $20 from 1 December13 Oct 2026
18Kimi K3.1 Pro Moonshot AI · open weights25.5%67.2%46$3.68$2 and $1024 Oct 2026
19Inkling 2 Small Thinking Machines · open weights25.4%13.1%31$1.18$0.30 and $15 Nov 2026
20GPT-6 Luna OpenAI · max effort25.2%12.6%28$0.15$0.10 and $0.5022 Sept 2026
  1. 1Claude Fable 5.2 Anthropic · max effort48.8%
    FrontierCode 2.0
    48.8%
    Terminal-⁠Bench 4.0
    59.6%
    Score
    65
    Cost per task
    $26.54
    Price per million tokens
    $8 and $40
    Released
    9 Nov 2026
  2. 2Claude Opus 5.5 Anthropic · max effort46.4%
    FrontierCode 2.0
    46.4%
    Terminal-⁠Bench 4.0
    59.6%
    Score
    59
    Cost per task
    $10.04
    Price per million tokens
    $4 and $20
    Released
    22 Sept 2026
  3. 3GPT-6 Astra OpenAI · max effort42.4%
    FrontierCode 2.0
    42.4%
    Terminal-⁠Bench 4.0
    59.1%
    Score
    57
    Cost per task
    $6.24
    Price per million tokens
    $10 and $50
    Released
    3 Sept 2026
  4. 4GPT-6.1 Terra OpenAI · max effort42.3%
    FrontierCode 2.0
    42.3%
    Terminal-⁠Bench 4.0
    55.6%
    Score
    51
    Cost per task
    $0.96
    Price per million tokens
    $1 and $5
    Released
    26 Oct 2026
  5. 5Claude Sonnet 5.5 Anthropic · max effort41.7%
    FrontierCode 2.0
    41.7%
    Terminal-⁠Bench 4.0
    63.6%
    Score
    55
    Cost per task
    $10.02
    Price per million tokens
    $2 and $10
    Released
    28 Sept 2026
  6. 6GPT-6.1 Sol OpenAI · max effort40.6%
    FrontierCode 2.0
    40.6%
    Terminal-⁠Bench 4.0
    56.1%
    Score
    55
    Cost per task
    $1.27
    Price per million tokens
    $2 and $10
    Released
    29 Sept 2026
  7. 7Gemini 4 Carbon Google · high effort40.6%
    FrontierCode 2.0
    40.6%
    Terminal-⁠Bench 4.0
    65.2%
    Score
    60
    Cost per task
    $5.48
    Price per million tokens
    $2 and $10 $4 and $20 from 1 December
    Released
    10 Nov 2026
  8. 8Gemini 4 Hydrogen Google · extra high effort39.0%
    FrontierCode 2.0
    39.0%
    Terminal-⁠Bench 4.0
    53.5%
    Score
    54
    Cost per task
    $3.16
    Price per million tokens
    $1.50 and $7.50
    Released
    12 Nov 2026
  9. 9Muse Wonder Full Meta · max effort37.8%
    FrontierCode 2.0
    37.8%
    Terminal-⁠Bench 4.0
    57.1%
    Score
    50
    Cost per task
    No data
    Price per million tokens
    No per-token price
    Released
    12 Oct 2026
  10. 10Claude Haiku 5.5 Anthropic · max effort34.9%
    FrontierCode 2.0
    34.9%
    Terminal-⁠Bench 4.0
    32.8%
    Score
    36
    Cost per task
    $0.41
    Price per million tokens
    $0.10 and $0.50
    Released
    7 Oct 2026
  11. 11Tencent Hy4 Tencent31.4%
    FrontierCode 2.0
    31.4%
    Terminal-⁠Bench 4.0
    37.4%
    Score
    35
    Cost per task
    $1.03
    Price per million tokens
    $0.834 and $2.501 half price until 31 December
    Released
    4 Nov 2026
  12. 12Grok 4.7 xAI · extra high effort28.7%
    FrontierCode 2.0
    28.7%
    Terminal-⁠Bench 4.0
    25.8%
    Score
    35
    Cost per task
    $7.77
    Price per million tokens
    $2 and $6
    Released
    21 Sept 2026
  13. 13DeepSeek V4.1 Flash 1017 DeepSeek · max effort · open weights28.5%
    FrontierCode 2.0
    28.5%
    Terminal-⁠Bench 4.0
    33.8%
    Score
    33
    Cost per task
    $0.60
    Price per million tokens
    $0.30 and $1.20
    Released
    17 Oct 2026
  14. 14Muse Wonder Meta · max effort27.9%
    FrontierCode 2.0
    27.9%
    Terminal-⁠Bench 4.0
    52.0%
    Score
    44
    Cost per task
    $6.95
    Price per million tokens
    $4 and $20
    Released
    12 Oct 2026
  15. 15Muse Ember Meta · max effort27.2%
    FrontierCode 2.0
    27.2%
    Terminal-⁠Bench 4.0
    40.9%
    Score
    36
    Cost per task
    $0.14
    Price per million tokens
    $0.10 and $0.50
    Released
    9 Nov 2026
  16. 16Qwen3.8-Flash-Next Alibaba · open weights27.1%
    FrontierCode 2.0
    27.1%
    Terminal-⁠Bench 4.0
    25.3%
    Score
    29
    Cost per task
    $0.57
    Price per million tokens
    No per-token price
    Released
    26 Aug 2026
  17. 17Gemini 4 Argon Google · high effort25.7%
    FrontierCode 2.0
    25.7%
    Terminal-⁠Bench 4.0
    57.1%
    Score
    50
    Cost per task
    $4.97
    Price per million tokens
    $2 and $10 $4 and $20 from 1 December
    Released
    13 Oct 2026
  18. 18Kimi K3.1 Pro Moonshot AI · open weights25.5%
    FrontierCode 2.0
    25.5%
    Terminal-⁠Bench 4.0
    67.2%
    Score
    46
    Cost per task
    $3.68
    Price per million tokens
    $2 and $10
    Released
    24 Oct 2026
  19. 19Inkling 2 Small Thinking Machines · open weights25.4%
    FrontierCode 2.0
    25.4%
    Terminal-⁠Bench 4.0
    13.1%
    Score
    31
    Cost per task
    $1.18
    Price per million tokens
    $0.30 and $1
    Released
    5 Nov 2026
  20. 20GPT-6 Luna OpenAI · max effort25.2%
    FrontierCode 2.0
    25.2%
    Terminal-⁠Bench 4.0
    12.6%
    Score
    28
    Cost per task
    $0.15
    Price per million tokens
    $0.10 and $0.50
    Released
    22 Sept 2026

FrontierCode 2.0 and Terminal-⁠Bench 4.0, the score and the cost per task from Artificial Analysis, read on 13 November 2026 with each model at its highest effort. The price and the release date from each maker's page. FrontierCode 2.0 and Terminal-⁠Bench 4.0 on Artificial Analysis.

Questions.

  • Which one for your company's software?

    No single one wins at everything. We explain it in the article “Which AI Model to Use to Build Your Company's Software”, including how we split the work between them.

  • What is the best AI for coding?

    By Artificial Analysis's tests on 13 November 2026, Claude Fable 5.2, from Anthropic, which solves 48.8% of FrontierCode, and Kimi K3.1 Pro, from Moonshot AI, with 67.2% of Terminal-⁠Bench. Neither wins at both.

  • What does FrontierCode measure?

    Whether the model can make a change to a real open-⁠source project that its maintainer would accept. It has 132 tasks, from Cognition, that Artificial Analysis runs the same way for every model.

  • What does Terminal-⁠Bench measure?

    Whether the model can finish complex work on its own in a computer's terminal, such as setting up, testing and fixing a program. It has 66 tasks in software, machine learning, science, operations, security and more.

  • What is the cost per task?

    What it costs on average for the model to do one of the tasks behind Artificial Analysis's score, across all its tests. It says more than the price per million tokens, because for the same work some models use many more tokens than others.

  • And the whole comparison?

    In “AI comparison 2026”, with every model ranked by its score, what each task costs and how long it takes, its price and its context.

Let's talk.

Tell us what you need and one of the partners who write the code gets back to you.