050%100%

RangeGPT-5.6 SolOpenAI · 3 values · 2 reports88.8%–91.9%median 88.8%RangeKimi K3Moonshot AI · 1 value · 1 report88.3%–88.3%median 88.3%RangeClaude Mythos 5Anthropic · 1 value · 1 report88.0%–88.0%median 88.0%RangeGPT-5.6 TerraOpenAI · 1 value · 1 report87.4%–87.4%median 87.4%RangeGPT-5.6 LunaOpenAI · 2 values · 2 reports84.7%–84.7%median 84.7%RangeClaude Fable 5Anthropic · 3 values · 3 reports83.1%–84.6%median 84.3%RangeGPT-5.5OpenAI · 6 values · 6 reports78.2%–85.6%median 83.4%RangeGrok 4.5SpaceXAI · 2 values · 2 reports83.3%–83.3%median 83.3%RangeClaude Opus 4.8Anthropic · 6 values · 6 reports78.9%–84.6%median 82.7%RangeGLM-5.2Z.ai · 1 value · 1 report82.7%–82.7%median 82.7%RangeClaude Sonnet 5Anthropic · 2 values · 2 reports80.4%–80.4%median 80.4%RangeMEMuse Spark 1.1Meta · 1 value · 1 report80.0%–80.0%median 80.0%RangeGemini 3.6 FlashGoogle · 1 value · 1 report78.0%–78.0%median 78.0%RangeGemini 3.5 FlashGoogle · 2 values · 2 reports76.2%–76.2%median 76.2%RangeClaude Opus 4.7Anthropic · 2 values · 2 reports66.1%–78.9%median 72.5%RangeGemini 3.1 Pro PreviewGoogle · 2 values · 2 reports70.7%–70.7%median 70.7%RangeGemini 3.1 ProGoogle · 3 values · 3 reports70.3%–73.8%median 70.3%RangeMEMuse Spark 1.0Meta · 1 value · 1 report67.3%–67.3%median 67.3%RangeClaude Sonnet 4.6Anthropic · 1 value · 1 report67.0%–67.0%median 67.0%RangeMiniMax M3MiniMax · 1 value · 1 report66.0%–66.0%median 66.0%RangeGemini 3 FlashGoogle · 1 value · 1 report58.0%–58.0%median 58.0%

Each row is one model. Sort order uses the published-value median; the line preserves its low-to-high range, and the dot marks the median—not an inferred score.