Qwen: Qwen3 235B A22B Thinking 2507

Description

Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-Experts (MoE) language model optimized for complex reasoning tasks. It activates 22B of its 235B parameters per forward pass and natively supports up to 262,144 tokens of context. This "thinking-only" variant enhances structured logical reasoning, mathematics, science, and long-form generation, showing strong benchmark performance across AIME, SuperGPQA, LiveCodeBench, and MMLU-Redux. It enforces a special reasoning mode (</think>) and is designed for high-token outputs (up to 81,920 tokens) in challenging domains. The model is instruction-tuned and excels at step-by-step reasoning, tool use, agentic workflows, and multilingual tasks. This release represents the most capable open-source variant in the Qwen3-235B series, surpassing many closed models in structured reasoning use cases.

Raw fields snapshot

{
  "id": "qwen/qwen3-235b-a22b-thinking-2507",
  "name": "Qwen: Qwen3 235B A22B Thinking 2507",
  "description": "Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-Experts (MoE) language model optimized for complex reasoning tasks. It activates 22B of its 235B parameters per forward pass and natively supports up to 262,144 tokens of context. This \"thinking-only\" variant enhances structured logical reasoning, mathematics, science, and long-form generation, showing strong benchmark performance across AIME, SuperGPQA, LiveCodeBench, and MMLU-Redux. It enforces a special reasoning mode (</think>) and is designed for high-token outputs (up to 81,920 tokens) in challenging domains.\n\nThe model is instruction-tuned and excels at step-by-step reasoning, tool use, agentic workflows, and multilingual tasks. This release represents the most capable open-source variant in the Qwen3-235B series, surpassing many closed models in structured reasoning use cases.",
  "created": 1753449557,
  "canonical_slug": "qwen/qwen3-235b-a22b-thinking-2507",
  "hugging_face_id": "Qwen/Qwen3-235B-A22B-Thinking-2507",
  "source_type": "both",
  "context_length": 131072,
  "max_completion_tokens": null,
  "is_moderated": false,
  "architecture": {
    "modality": "text->text",
    "input_modalities": [
      "text"
    ],
    "output_modalities": [
      "text"
    ],
    "tokenizer": "Qwen3",
    "instruct_type": "qwen3"
  },
  "input_modalities": [
    "text"
  ],
  "output_modalities": [
    "text"
  ],
  "modality": "text->text",
  "tokenizer": "Qwen3",
  "instruct_type": "qwen3",
  "supported_parameters": [
    "frequency_penalty",
    "include_reasoning",
    "logit_bias",
    "max_tokens",
    "min_p",
    "presence_penalty",
    "reasoning",
    "repetition_penalty",
    "response_format",
    "seed",
    "stop",
    "structured_outputs",
    "temperature",
    "tool_choice",
    "tools",
    "top_k",
    "top_p"
  ],
  "default_parameters": {
    "temperature": null,
    "top_p": null,
    "frequency_penalty": null
  },
  "per_request_limits": null,
  "top_provider": {
    "context_length": 131072,
    "max_completion_tokens": null,
    "is_moderated": false
  },
  "pricing": {
    "prompt": "0",
    "completion": "0",
    "request": "0",
    "image": "0",
    "web_search": "0",
    "internal_reasoning": "0"
  },
  "PPM": {
    "prompt": 0,
    "completion": 0,
    "request": 0,
    "image": 0,
    "web_search": 0,
    "internal_reasoning": 0
  },
  "openrouter_raw": {
    "id": "qwen/qwen3-235b-a22b-thinking-2507",
    "canonical_slug": "qwen/qwen3-235b-a22b-thinking-2507",
    "hugging_face_id": "Qwen/Qwen3-235B-A22B-Thinking-2507",
    "name": "Qwen: Qwen3 235B A22B Thinking 2507",
    "created": 1753449557,
    "description": "Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-Experts (MoE) language model optimized for complex reasoning tasks. It activates 22B of its 235B parameters per forward pass and natively supports up to 262,144 tokens of context. This \"thinking-only\" variant enhances structured logical reasoning, mathematics, science, and long-form generation, showing strong benchmark performance across AIME, SuperGPQA, LiveCodeBench, and MMLU-Redux. It enforces a special reasoning mode (</think>) and is designed for high-token outputs (up to 81,920 tokens) in challenging domains.\n\nThe model is instruction-tuned and excels at step-by-step reasoning, tool use, agentic workflows, and multilingual tasks. This release represents the most capable open-source variant in the Qwen3-235B series, surpassing many closed models in structured reasoning use cases.",
    "context_length": 131072,
    "architecture": {
      "modality": "text->text",
      "input_modalities": [
        "text"
      ],
      "output_modalities": [
        "text"
      ],
      "tokenizer": "Qwen3",
      "instruct_type": "qwen3"
    },
    "pricing": {
      "prompt": "0",
      "completion": "0",
      "request": "0",
      "image": "0",
      "web_search": "0",
      "internal_reasoning": "0"
    },
    "top_provider": {
      "context_length": 131072,
      "max_completion_tokens": null,
      "is_moderated": false
    },
    "per_request_limits": null,
    "supported_parameters": [
      "frequency_penalty",
      "include_reasoning",
      "logit_bias",
      "max_tokens",
      "min_p",
      "presence_penalty",
      "reasoning",
      "repetition_penalty",
      "response_format",
      "seed",
      "stop",
      "structured_outputs",
      "temperature",
      "tool_choice",
      "tools",
      "top_k",
      "top_p"
    ],
    "default_parameters": {
      "temperature": null,
      "top_p": null,
      "frequency_penalty": null
    },
    "expiration_date": null
  },
  "categories": {
    "text-expert": {
      "score": 1465,
      "rank_ub": 25,
      "votes": 417,
      "ci_95": "±29"
    },
    "text-industry-software-and-it-services": {
      "score": 1431,
      "rank_ub": 74,
      "votes": 3035,
      "ci_95": "±11"
    },
    "text-industry-writing-and-literature-and-language": {
      "score": 1377,
      "rank_ub": 76,
      "votes": 1932,
      "ci_95": "±13"
    },
    "text-overall": {
      "score": 1399,
      "rank_ub": 75,
      "votes": 9186,
      "ci_95": "±6"
    },
    "text-industry-life-and-physical-and-social-science": {
      "score": 1411,
      "rank_ub": 80,
      "votes": 1544,
      "ci_95": "±15"
    },
    "text-industry-mathematical": {
      "score": 1430,
      "rank_ub": 38,
      "votes": 508,
      "ci_95": "±25"
    },
    "text-industry-entertainment-and-sports-and-media": {
      "score": 1351,
      "rank_ub": 90,
      "votes": 1600,
      "ci_95": "±15"
    },
    "text-industry-business-and-management-and-financial-operations": {
      "score": 1399,
      "rank_ub": 70,
      "votes": 1569,
      "ci_95": "±15"
    },
    "text-industry-medicine-and-healthcare": {
      "score": 1434,
      "rank_ub": 68,
      "votes": 533,
      "ci_95": "±26"
    },
    "text-industry-legal-and-government": {
      "score": 1388,
      "rank_ub": 94,
      "votes": 561,
      "ci_95": "±24"
    },
    "text-instruction-following": {
      "score": 1388,
      "rank_ub": 74,
      "votes": 2140,
      "ci_95": "±12"
    },
    "text-math": {
      "score": 1407,
      "rank_ub": 68,
      "votes": 505,
      "ci_95": "±24"
    },
    "text-creative-writing": {
      "score": 1372,
      "rank_ub": 75,
      "votes": 1127,
      "ci_95": "±18"
    },
    "text-multi-turn": {
      "score": 1393,
      "rank_ub": 81,
      "votes": 1416,
      "ci_95": "±15"
    },
    "text-hard-prompts": {
      "score": 1418,
      "rank_ub": 74,
      "votes": 3912,
      "ci_95": "±9"
    },
    "text-coding": {
      "score": 1441,
      "rank_ub": 75,
      "votes": 1625,
      "ci_95": "±15"
    },
    "text-hard-prompts-english": {
      "score": 1427,
      "rank_ub": 71,
      "votes": 2054,
      "ci_95": "±13"
    },
    "text-longer-query": {
      "score": 1402,
      "rank_ub": 80,
      "votes": 1667,
      "ci_95": "±14"
    },
    "text-english": {
      "score": 1405,
      "rank_ub": 83,
      "votes": 4478,
      "ci_95": "±9"
    },
    "text-chinese": {
      "score": 1452,
      "rank_ub": 45,
      "votes": 425,
      "ci_95": "±29"
    },
    "text-spanish": {
      "score": 1373,
      "rank_ub": 87,
      "votes": 155,
      "ci_95": "±44"
    },
    "text-russian": {
      "score": 1400,
      "rank_ub": 64,
      "votes": 440,
      "ci_95": "±26"
    },
    "text-german": {
      "score": 1391,
      "rank_ub": 48,
      "votes": 209,
      "ci_95": "±40"
    },
    "text-japanese": {
      "score": 1363,
      "rank_ub": 45,
      "votes": 432,
      "ci_95": "±30"
    },
    "text-korean": {
      "score": 1340,
      "rank_ub": 62,
      "votes": 203,
      "ci_95": "±42"
    },
    "text-exclude-ties": {
      "score": 1380,
      "rank_ub": 75,
      "votes": 6473,
      "ci_95": "±9"
    }
  },
  "arena_model_id": "qwen3-235b-a22b-thinking-2507",
  "leaderboard_name": "qwen3-235b-a22b-thinking-2507",
  "match_method": "openrouter_name",
  "match_key": "qwen3-235b-a22b-thinking-2507",
  "match_input": "Qwen: Qwen3 235B A22B Thinking 2507",
  "arena_aliases": []
}