Coding Agents

Which Artificial Analysis coding agent and model pairing gives you the most intelligence per dollar — and where you are overpaying for the same score.

Best Models

The most cost-efficient model for each performance score, from all available models.

  1. Muse Code - Muse Spark 1.3 (max)· Meta
    68.0%
    $0.023/pt
  2. Codex - GPT-5.6 Luna (max)· 82% cheaper · 10.8% worse than Muse Code - Muse Spark 1.3 (max)
    57.2%
    0.504¢/pt
  3. Codex - GPT-5.6 Luna (high)· 38% cheaper · 5.5% worse than Codex - GPT-5.6 Luna (max)
    51.7%
    0.348¢/pt
  4. Codex - DeepSeek V4 Flash 0731 (max)· 34% cheaper · 1.9% worse than Codex - GPT-5.6 Luna (high)
    49.8%
    0.239¢/pt
  5. Cursor CLI - Composer 2.5· 29% cheaper · 11.5% worse than Codex - DeepSeek V4 Flash 0731 (max)
    38.3%
    0.222¢/pt
  6. Codex - GPT-5.6 Luna (low)· 54% cheaper · 13.2% worse than Cursor CLI - Composer 2.5
    25.1%
    0.155¢/pt
PerformanceCost ($1 blocks, 10¢ blocks)Empty

Models in the same performance bracket are treated as similar-score alternatives in the comparison.

Charts

Compare artificialanalysis.ai pass rate with mean cost. The top-right corner is the sweet spot: higher performance for less money.

68 selected models
Providers

Chart loads in the browser.

Comparison Table

68 models · sort by performance or price

Sort comparison table
  • Claude Code - Fable 5.1 (max) (with fallback)· Anthropic
    70.4%
    $0.130/pt
  • Claude Code - Opus 5 (xhigh)· 11% cheaper · 2.3% worse than Claude Code - Fable 5.1 (max) (with fallback)
    68.1%
    $0.120/pt
  • Muse Code - Muse Spark 1.3 (max)· 81% cheaper · 0.2% worse than Claude Code - Opus 5 (xhigh)
    68.0%
    $0.023/pt
  • Claude Code - Fable 5 (max) (with fallback)· 638% pricier · 0.8% worse than Muse Code - Muse Spark 1.3 (max)
    67.2%
    $0.174/pt
  • Claude Code - Opus 5 (max)· 23% cheaper · 0.1% worse than Claude Code - Fable 5 (max) (with fallback)
    67.0%
    $0.133/pt
  • Codex - GPT-6 Astra (max)· 47% cheaper · 0.1% worse than Claude Code - Opus 5 (max)
    67.0%
    $0.070/pt
  • Claude Code - Fable 5 (xhigh) (with fallback)· 81% pricier · 1.0% worse than Codex - GPT-6 Astra (max)
    66.0%
    $0.129/pt
  • Claude Code - Opus 5 (high)· 54% cheaper · 0.4% worse than Claude Code - Fable 5 (xhigh) (with fallback)
    65.6%
    $0.060/pt
  • Claude Code - Fable 5 (high) (with fallback)· 52% pricier · 0.5% worse than Claude Code - Opus 5 (high)
    65.1%
    $0.092/pt
  • Codex - GPT-5.6 Sol (max)· 16% cheaper · 0.1% worse than Claude Code - Fable 5 (high) (with fallback)
    65.1%
    $0.077/pt
  • Muse Code - Muse Spark 1.3 (xhigh)· 68% cheaper · 0.9% worse than Codex - GPT-5.6 Sol (max)
    64.2%
    $0.025/pt
  • Codex - GPT-5.6 Sol (high)· 86% pricier · 0.1% worse than Muse Code - Muse Spark 1.3 (xhigh)
    64.1%
    $0.047/pt
  • Grok Build - Grok 4.5 (high)· 19% cheaper · 0.0% worse than Codex - GPT-5.6 Sol (high)
    64.1%
    $0.038/pt
  • Claude Code - Opus 5 (medium)· 30% pricier · 0.0% worse than Grok Build - Grok 4.5 (high)
    64.1%
    $0.050/pt
  • Codex - GPT-5.6 Sol (xhigh)· 18% pricier · 0.7% worse than Claude Code - Opus 5 (medium)
    63.3%
    $0.059/pt
  • Claude Code - Fable 5 (medium) (with fallback)· 27% pricier · 0.1% worse than Codex - GPT-5.6 Sol (xhigh)
    63.2%
    $0.075/pt
  • Kimi Code CLI - Kimi K3· 35% cheaper · 0.6% worse than Claude Code - Fable 5 (medium) (with fallback)
    62.6%
    $0.049/pt
  • Claude Code - Opus 4.8 (max)· 151% pricier · 0.5% worse than Kimi Code CLI - Kimi K3
    62.1%
    $0.124/pt
  • Muse Code - Muse Spark 1.2 (xhigh)· 73% cheaper · 0.5% worse than Claude Code - Opus 4.8 (max)
    61.6%
    $0.034/pt
  • Codex - GPT-5.6 Sol (medium)· 6% pricier · 0.0% worse than Muse Code - Muse Spark 1.2 (xhigh)
    61.6%
    $0.036/pt
  • Claude Code - Qwen3.8 Max· 47% pricier · 0.3% worse than Codex - GPT-5.6 Sol (medium)
    61.3%
    $0.053/pt
  • Opencode - Gemini 3.8 Flash (high)· 37% cheaper · 0.2% worse than Claude Code - Qwen3.8 Max
    61.2%
    $0.033/pt
  • Codex - GPT-5.5 (xhigh)· 133% pricier · 0.1% worse than Opencode - Gemini 3.8 Flash (high)
    61.0%
    $0.078/pt
  • Codex - GPT-5.6 Terra (max)· 59% cheaper · 0.6% worse than Codex - GPT-5.5 (xhigh)
    60.4%
    $0.032/pt
  • Opencode - Gemini 3.7 Flash (high)· 34% cheaper · 0.8% worse than Codex - GPT-5.6 Terra (max)
    59.6%
    $0.021/pt
  • Claude Code - Opus 5 (low)· 81% pricier · 0.2% worse than Opencode - Gemini 3.7 Flash (high)
    59.4%
    $0.039/pt
  • Claude Code - Opus 5 (none)· 54% pricier · 0.2% worse than Claude Code - Opus 5 (low)
    59.2%
    $0.060/pt
  • Antigravity SDK - Gemini 3.8 Flash (high)· 43% cheaper · 0.1% worse than Claude Code - Opus 5 (none)
    59.1%
    $0.034/pt
  • Claude Code - Opus 4.8 (xhigh)· 182% pricier · 0.0% worse than Antigravity SDK - Gemini 3.8 Flash (high)
    59.1%
    $0.096/pt
  • Opencode - Muse Spark 1.2 (xhigh)· 66% cheaper · 0.2% worse than Claude Code - Opus 4.8 (xhigh)
    58.9%
    $0.033/pt
  • Claude Code - Fable 5 (low) (with fallback)· 65% pricier · 0.3% worse than Opencode - Muse Spark 1.2 (xhigh)
    58.6%
    $0.054/pt
  • Claude Code - Opus 4.8 (high)· 19% pricier · 1.0% worse than Claude Code - Fable 5 (low) (with fallback)
    57.6%
    $0.066/pt
  • Codex - GPT-5.6 Luna (max)· 92% cheaper · 0.4% worse than Claude Code - Opus 4.8 (high)
    57.2%
    0.504¢/pt
  • Antigravity SDK - Gemini 3.7 Flash (high)· 384% pricier · 0.6% worse than Codex - GPT-5.6 Luna (max)
    56.6%
    $0.025/pt
  • Codex - GPT-5.6 Terra (xhigh)· 3% cheaper · 0.5% worse than Antigravity SDK - Gemini 3.7 Flash (high)
    56.0%
    $0.024/pt
  • Claude Code - Opus 4.8 (medium)· 143% pricier · 0.3% worse than Codex - GPT-5.6 Terra (xhigh)
    55.8%
    $0.059/pt
  • Codex - GPT-5.5 (medium)· 20% cheaper · 0.5% worse than Claude Code - Opus 4.8 (medium)
    55.3%
    $0.048/pt
  • Codex - GPT-5.6 Sol (low)· 51% cheaper · 0.1% worse than Codex - GPT-5.5 (medium)
    55.2%
    $0.023/pt
  • Opencode - Muse Spark 1.1 (xhigh)· 11% pricier · 0.3% worse than Codex - GPT-5.6 Sol (low)
    54.9%
    $0.026/pt
  • Codex - GPT-5.6 Terra (high)· 21% cheaper · 0.3% worse than Opencode - Muse Spark 1.1 (xhigh)
    54.6%
    $0.021/pt
  • Codex - GPT-5.6 Luna (xhigh)· 79% cheaper · 1.7% worse than Codex - GPT-5.6 Terra (high)
    53.0%
    0.445¢/pt
  • Devin CLI - SWE-1.7 Lightning Max· 3512% pricier · 0.7% worse than Codex - GPT-5.6 Luna (xhigh)
    52.2%
    $0.163/pt
  • Codex - GPT-5.6 Luna (high)· 98% cheaper · 0.5% worse than Devin CLI - SWE-1.7 Lightning Max
    51.7%
    0.348¢/pt
  • Claude Code - Opus 4.7 (max)· 3190% pricier · 0.1% worse than Codex - GPT-5.6 Luna (high)
    51.6%
    $0.115/pt
  • Opencode - Opus 4.7 (medium)· 50% cheaper · 0.3% worse than Claude Code - Opus 4.7 (max)
    51.3%
    $0.057/pt
  • Codex - DeepSeek V4 Flash 0731 (max)· 96% cheaper · 1.5% worse than Opencode - Opus 4.7 (medium)
    49.8%
    0.239¢/pt
  • Claude Code - Opus 4.8 (low)· 1734% pricier · 0.7% worse than Codex - DeepSeek V4 Flash 0731 (max)
    49.0%
    $0.045/pt
  • Codex - GPT-5.6 Terra (medium)· 69% cheaper · 1.0% worse than Claude Code - Opus 4.8 (low)
    48.0%
    $0.014/pt
  • Opencode - Gemini 3.6 Flash (high)· 211% pricier · 0.8% worse than Codex - GPT-5.6 Terra (medium)
    47.2%
    $0.044/pt
  • Cursor CLI - GPT-5.5 (medium)· 4% cheaper · 0.2% worse than Opencode - Gemini 3.6 Flash (high)
    47.1%
    $0.042/pt
  • Cursor CLI - Opus 4.7 (medium)· 36% pricier · 0.4% worse than Cursor CLI - GPT-5.5 (medium)
    46.7%
    $0.058/pt
  • Codex - GPT-5.6 Sol (none)· 60% cheaper · 3.3% worse than Cursor CLI - Opus 4.7 (medium)
    43.4%
    $0.025/pt
  • Claude Code - GLM-5.2· 76% pricier · 0.1% worse than Codex - GPT-5.6 Sol (none)
    43.3%
    $0.044/pt
  • Codex - DeepSeek V4 Pro 0813 (max)· 90% cheaper · 0.5% worse than Claude Code - GLM-5.2
    42.8%
    0.425¢/pt
  • Claude Code - Opus 4.7 (medium)· 891% pricier · 0.4% worse than Codex - DeepSeek V4 Pro 0813 (max)
    42.4%
    $0.042/pt
  • Codex - GPT-5.6 Luna (medium)· 95% cheaper · 0.4% worse than Claude Code - Opus 4.7 (medium)
    42.0%
    0.211¢/pt
  • Claude Code - Sonnet 4.6 (medium)· 2209% pricier · 3.3% worse than Codex - GPT-5.6 Luna (medium)
    38.6%
    $0.053/pt
  • Codex - GPT-5.6 Terra (low)· 81% cheaper · 0.0% worse than Claude Code - Sonnet 4.6 (medium)
    38.6%
    0.996¢/pt
  • Cursor CLI - Composer 2.5· 78% cheaper · 0.3% worse than Codex - GPT-5.6 Terra (low)
    38.3%
    0.222¢/pt
  • Cursor CLI - Composer 2.5 Fast· 555% pricier · 0.0% better than Cursor CLI - Composer 2.5
    38.3%
    $0.015/pt
  • Claude Code - Qwen3.7 Plus (thinking)· 1030% pricier · 0.2% worse than Cursor CLI - Composer 2.5 Fast
    38.1%
    $0.165/pt
  • Claude Code - GLM-5.1· 32% cheaper · 1.2% worse than Claude Code - Qwen3.7 Plus (thinking)
    36.9%
    $0.116/pt
  • Claude Code - Kimi K2.6· 72% cheaper · 3.0% worse than Claude Code - GLM-5.1
    33.9%
    $0.036/pt
  • Gemini CLI - Gemini 3.1 Pro (high)· 15% cheaper · 1.0% worse than Claude Code - Kimi K2.6
    32.9%
    $0.032/pt
  • Claude Code - DeepSeek V4 Pro (high)· 51% cheaper · 0.1% worse than Gemini CLI - Gemini 3.1 Pro (high)
    32.8%
    $0.016/pt
  • Codex - GPT-5.6 Luna (low)· 92% cheaper · 7.8% worse than Claude Code - DeepSeek V4 Pro (high)
    25.1%
    0.155¢/pt
  • Codex - GPT-5.6 Terra (none)· 652% pricier · 2.0% worse than Codex - GPT-5.6 Luna (low)
    23.1%
    $0.013/pt
  • Codex - GPT-5.6 Luna (none)· 76% cheaper · 4.0% worse than Codex - GPT-5.6 Terra (none)
    19.1%
    0.364¢/pt
PerformanceCost ($1 blocks, 10¢ blocks)Empty

Models in the same performance bracket are treated as similar-score alternatives in the comparison.

Model selector

Choose the model and reasoning effort rows to compare.

68 of 68 selected
ProvidersAll providers shown

Alibaba

Claude Code - Qwen3.8 Max

Alibaba

Claude Code - Qwen3.7 Plus

Alibaba

Anthropic

Claude Code - Fable 5.1 (with fallback)

Anthropic

Claude Code - Opus 5

Anthropic

Claude Code - Fable 5 (with fallback)

Anthropic

Claude Code - Opus 4.8

Anthropic

Claude Code - Opus 4.7

Anthropic

Opencode - Opus 4.7

Anthropic

Claude Code - Sonnet 4.6

Anthropic

DeepSeek

Codex - DeepSeek V4 Flash 0731

DeepSeek

Codex - DeepSeek V4 Pro 0813

DeepSeek

Claude Code - DeepSeek V4 Pro

DeepSeek

Google

Opencode - Gemini 3.8 Flash

Google

Opencode - Gemini 3.7 Flash

Google

Antigravity SDK v0.1.12 - Gemini 3.8 Flash

Google

Antigravity SDK v0.1.8 - Gemini 3.7 Flash

Google

Opencode - Gemini 3.6 Flash

Google

Gemini CLI - Gemini 3.1 Pro

Google

Meta

Muse Code 1.0.2 RC - Muse Spark 1.3

Meta

Muse Code - Muse Spark 1.2

Meta

Opencode - Muse Spark 1.2

Meta

Opencode - Muse Spark 1.1

Meta

Moonshot

Kimi Code CLI - Kimi K3

Moonshot

Claude Code - Kimi K2.6

Moonshot

OpenAI

Codex - GPT-6 Astra

OpenAI

Codex - GPT-5.6 Sol

OpenAI

Codex - GPT-5.5

OpenAI

Codex - GPT-5.6 Terra

OpenAI

Codex - GPT-5.6 Luna

OpenAI

Other

Devin CLI - SWE-1.7 Lightning Max

Other

xAI

Grok Build - Grok 4.5

xAI

Cursor CLI - GPT-5.5

xAI

Cursor CLI - Opus 4.7

xAI

Cursor CLI - Composer 2.5

xAI

Cursor CLI - Composer 2.5 Fast

xAI

Zhipu

Claude Code - GLM-5.2

Zhipu

Claude Code - GLM-5.1

Zhipu