moerew's picture
Publish Best LLM for Coding: The Arena Coding Leaderboard benchmark dataset
b2df220 verified
Raw History Blame Contribute Delete
2.14 kB
{
"schema": "modevc.chart.v1",
"slug": "best-llm-for-coding",
"title": "Best LLM for Coding: The Arena Coding Leaderboard",
"date": "2026-09-09",
"canonical_url": "https://modevc.com/charts/best-llm-for-coding/",
"claim": "claude-opus-4-6-high leads Arena coding at 1535. Anthropic has held #1 for 280 days, but Z.ai and Google sit within the interval, 19 points back.",
"takeaway": "In coding, the lead has lasted 9 months. The challenger is already inside the margin of error.",
"sources": [
{
"name": "Arena leaderboard dataset (lmarena-ai, CC BY 4.0)",
"url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"
},
{
"name": "Z.ai GLM-5.3-Flash release note",
"url": "https://docs.z.ai/guides/vlm/glm-5.3-flash"
}
],
"data": {
"type": "hbar",
"fmt": "pts",
"labelHead": "Best model (lab)",
"valueHead": "Points above #50",
"rows": [
{
"label": "claude-opus-4-6-high (Anthropic)",
"v": 61.5,
"logo": "si-anthropic"
},
{
"label": "glm-5.3-flash (Z.ai)",
"v": 42.2,
"logo": "lh-zai"
},
{
"label": "gemini-3.8-flash-high (Google)",
"v": 42.1,
"logo": "si-google"
},
{
"label": "kimi-k3-max (Moonshot)",
"v": 36.7,
"logo": "lh-moonshot"
},
{
"label": "mimo-v2.5-pro (Xiaomi)",
"v": 26.6,
"logo": "si-xiaomi"
},
{
"label": "qwen3.8-max (Alibaba)",
"v": 24.5,
"logo": "si-qwen"
},
{
"label": "muse-spark-1.2 (xHigh) (Meta)",
"v": 22.3,
"logo": "si-meta"
},
{
"label": "gpt-5.4-high (OpenAI)",
"v": 21.5,
"logo": "lh-openai"
},
{
"label": "ernie-5.1 (Baidu)",
"v": 14.9,
"logo": "lh-baidu"
},
{
"label": "amazon-nova-experimental-chat-26-02-10 (Amazon)",
"v": 11.2,
"logo": "lh-aws"
},
{
"label": "grok-4.5 (xAI)",
"v": 8.7,
"logo": "lh-xai"
}
]
}
}