Qwen3.8 27B

{
  "id": "alibaba/qwen3.8-27b",
  "name": "Qwen3.8 27B",
  "description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
  "family": "qwen",
  "last_updated": "2026-08-14",
  "attachment": true,
  "reasoning": true,
  "tool_call": true,
  "structured_output": true,
  "temperature": true,
  "modalities": {
    "input": [
      "text",
      "image",
      "video"
    ],
    "output": [
      "text"
    ]
  },
  "limit": {
    "context": 262144,
    "output": 32768
  },
  "benchmarks": [
    {
      "name": "SWE-bench Pro",
      "score": 61.7,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "resolved"
    },
    {
      "name": "Terminal-Bench",
      "score": 73,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "harness": "Terminus",
      "version": "2.1"
    },
    {
      "name": "NL2Repo",
      "score": 42.3,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "harness": "Claude Code"
    },
    {
      "name": "DeepSWE",
      "score": 42.2,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "harness": "Claude Code",
      "version": "1.1"
    },
    {
      "name": "QwenSWEBench",
      "score": 79,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "avg@3",
      "harness": "Claude Code",
      "dataset": "internal"
    },
    {
      "name": "CoWorkBench",
      "score": 70.7,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "dataset": "internal"
    },
    {
      "name": "JobBench",
      "score": 33.4,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score"
    },
    {
      "name": "Agents' Last Exam",
      "score": 20.4,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "pass@1"
    },
    {
      "name": "Agents' Last Exam",
      "score": 42.9,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score"
    },
    {
      "name": "IFBench",
      "score": 79.5,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score"
    },
    {
      "name": "GPQA Diamond",
      "score": 89.2,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score"
    },
    {
      "name": "Humanity's Last Exam",
      "score": 30.8,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "variant": "without tools; GPT-4o judge"
    },
    {
      "name": "LiveCodeBench",
      "score": 90.3,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "version": "6"
    },
    {
      "name": "OSWorld-Verified",
      "score": 84.3,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score"
    },
    {
      "name": "WebArena-Verified",
      "score": 64.8,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "harness": "OSWorld scaffold; official WebArena-Verified grader"
    },
    {
      "name": "AndroidWorld",
      "score": 81.9,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score"
    },
    {
      "name": "SWE-MM",
      "score": 38.6,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score"
    },
    {
      "name": "RealWorldQA",
      "score": 85.9,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score"
    },
    {
      "name": "ERQA",
      "score": 65.5,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score"
    },
    {
      "name": "RecreationBench",
      "score": 47.1,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "dataset": "internal"
    },
    {
      "name": "ClawEval-MM",
      "score": 57.4,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "pass@3"
    },
    {
      "name": "ClawEval-MM",
      "score": 56.9,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "average score"
    },
    {
      "name": "Vision2Web",
      "score": 62.9,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "harness": "Claude Code",
      "variant": "gpt-5.4-2026-03-05 judge"
    },
    {
      "name": "MathVision",
      "score": 90,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "variant": "without code interpreter",
      "dataset": "corrected annotations"
    },
    {
      "name": "MathVision",
      "score": 94.6,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "variant": "with code interpreter",
      "dataset": "corrected annotations"
    },
    {
      "name": "BabyVision",
      "score": 65.7,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "variant": "without code interpreter"
    },
    {
      "name": "BabyVision",
      "score": 85.6,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "variant": "with code interpreter"
    },
    {
      "name": "CharXiv Reasoning",
      "score": 83.7,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "variant": "without code interpreter",
      "dataset": "corrected annotations"
    },
    {
      "name": "CharXiv Reasoning",
      "score": 90.2,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "variant": "with code interpreter",
      "dataset": "corrected annotations"
    },
    {
      "name": "OmniDocBench",
      "score": 91.1,
      "source": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "metric": "score",
      "version": "1.5"
    }
  ],
  "release_date": "2026-08-14",
  "open_weights": true,
  "weights": [
    {
      "label": "Hugging Face",
      "url": "https://huggingface.co/Qwen/Qwen3.8-27B"
    }
  ]
}