{
  "updated": "2026-08-30",
  "note": "Measured against the live gateway. More accurate than the vendor console.",
  "tiers": {
    "team": {
      "chat_models": 18
    },
    "personal": {
      "chat_models": 9
    }
  },
  "endpoints": {
    "global": "https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1",
    "cn": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1"
  },
  "anthropic_endpoints": {
    "global": "https://token-plan.ap-southeast-1.maas.aliyuncs.com/apps/anthropic",
    "cn": "https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic"
  },
  "capability_notes": {
    "capability_scope": "Capability values in this file were measured against the OpenAI-compatible surface (/compatible-mode/v1) unless noted. structured_output here means response_format json_schema; thinking here means enable_thinking + reasoning_content. The Anthropic-compatible surface (/apps/anthropic) was SEPARATELY tested 2026-07-24 - see meta.anthropic_surface. It has no /models listing (404) but every model is callable by id. Matters for Claude Code users, who use that surface.",
    "anthropic_surface": "TESTED 2026-07-24 (Global/Singapore, POST /apps/anthropic/v1/messages, x-api-key + anthropic-version:2023-06-01). FULLY FUNCTIONAL. All 15 chat models callable by the same id (no /models listing - 404 - but direct calls work). Returns NATIVE Anthropic format: thinking + text content blocks, stop_reason, and real cache accounting (cache_creation_input_tokens / cache_read_input_tokens). The thinking param (thinking:{type:enabled,budget_tokens}) is accepted on all 15. Native tool_use (auto tool_choice) works on all 15. Vision via Anthropic image blocks works on the same 7 multimodal models as the OpenAI surface (qwen3.8-max-preview, qwen3.7-plus, qwen3.6-plus, qwen3.6-flash, kimi-k2.5/k2.6/k2.7-code). FORCED tool_choice {type:tool} (the Anthropic structured-output path) works on deepseek-v3.2, kimi-k2.5/k2.6/k2.7-code, glm-5/glm-5.1/glm-5.2 (7) but is REJECTED with 400 by all 5 Qwen, deepseek-v4-pro, deepseek-v4-flash and MiniMax-M2.5 (8). This DIFFERS from the OpenAI surface (where tool_choice:required was honoured by none), so forced-tool structured output is surface-specific.",
    "openai_compatibility": "OpenAI-param support on /chat/completions, gateway-verified 2026-07-24. UNIVERSAL (all 15 chat models): top_p, frequency_penalty, presence_penalty, stop, seed, max_completion_tokens, streaming. logprobs: per-model (capabilities.logprobs) - Qwen, Kimi, glm-5.1/glm-5 yes; deepseek, MiniMax, glm-5.2 no. n>1 (multiple choices): only where thinking can be OFF (deepseek-v3.2, kimi-k2.5, kimi-k2.6); every other model returns 'n must be 1 when enable_thinking is true'. tool_choice: only 'auto' is honoured - 'required' is REJECTED by Qwen (400) and silently ignored by the rest, so a forced tool call cannot be relied on anywhere. Parallel tool calls DO work (multiple tool_calls returned in one response).",
    "knowledge_cutoff_note": "NOT recorded per model: self-reported cutoffs (asked each model directly, 2026-07-24) are unreliable and self-contradictory - the Qwen models claim '2026' while sibling qwen3.6-flash said 'July 2024'; others ranged Jan-2024 to Jan-2025. Do not treat model self-report as authoritative. Source from vendor model cards if needed."
  },
  "models": [
    {
      "id": "qwen3.8-max",
      "vendor": "Qwen",
      "kind": "chat",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-08-03",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 131072
      },
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": true,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": true,
        "vision_tool_result": true,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 58,
      "aa_url": "https://artificialanalysis.ai/models/qwen3-8-max",
      "cost": {
        "status": "published per-token pricing (GA), CN-denominated sheet",
        "per_1m_rmb": {
          "input": 12,
          "output": 36,
          "cache_hit": 1.5
        },
        "note": "From the qianwenai.com marketplace card (2026-08-04): input \u00a512/M, output \u00a536/M, cache-hit \u00a51.5/M, explicit-cache \u00a515 write / \u00a51 read per M. The preview's '10% of standard' promotion resolves against this standard rate. Reasoning tokens bill as output (always-on overhead no longer applies since thinking can be disabled).",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 12,
        "output": 36,
        "cache_hit": 1.5,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-04"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "reasoning_effort": {
          "accepted": [
            "low"
          ],
          "documented": [
            "low",
            "high",
            "xhigh"
          ],
          "default": "xhigh",
          "note": "'low' gateway-verified behaviourally (2026-08-05). 'medium' now acceptance-verified on the GA id via the validation enum (2026-08-05); behavioural response per level still untested. Default xhigh.",
          "validation_enum": [
            "none",
            "minimal",
            "low",
            "medium",
            "high",
            "xhigh",
            "max"
          ],
          "enum_note": "Gateway validation acceptance enumerated 2026-08-05 via zero-cost invalid-value probe. Acceptance, not behaviour \u2014 the lists are partly a generic superset (2026-07-24 sweep); per-model differences here match July's behavioural findings."
        },
        "budget_max": 262144
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen3.8-max",
        "cn": "https://www.qianwenai.com/models/qwen3.8-max"
      }
    },
    {
      "id": "qwen3.8-flash",
      "vendor": "Qwen",
      "kind": "chat",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-08-26",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 131072
      },
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": true,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": false,
        "vision_tool_result": true,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": false
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 0.8,
          "output": 2.7,
          "cache_hit": 0.1
        },
        "note": "CN card (qianwenai.com, 2026-08-30, authoritative per the marketplace-card rule): input CNY0.8/M, output CNY2.7/M, cache-hit CNY0.1/M, explicit-cache create CNY1.25/M / read CNY0.1/M. EN twin (qwencloud.com): $0.15 / $0.47 / $0.016 / $0.2 / $0.016 \u2014 EN is ~40% dearer at FX; the CN sheet is what meters Token Plan credits. Card shows batch + fine-tune flags \u2014 those describe Model Studio platform services, NOT Token Plan gateway routes (batch/fine-tuning routes 404 on the Global gateway, proven 2026-08-04).",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 0.8,
        "output": 2.7,
        "cache_hit": 0.1,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-30"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 262144,
        "note": "Card-documented: thinking max 983K input tokens / 131K output, max reasoning tokens 262K. Levels NOT enumerated or probed yet (spend gate) \u2014 inherited from qwen3.8 family pattern, not measured.",
        "effort": true
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen3.8-flash",
        "cn": "https://www.qianwenai.com/models/qwen3.8-flash"
      }
    },
    {
      "id": "qwen3.7-max",
      "vendor": "Qwen",
      "kind": "chat",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-05-21",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 131072
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": true,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 47,
      "aa_url": "https://artificialanalysis.ai/models/qwen3-7-max",
      "status_note": "AA Intelligence Index page marks this model deprecated/superseded (by qwen3.8-max) as of 2026-08-18; gateway still serves it (status active here).",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet (limited-time 50% off promo active)",
        "per_1m_rmb": {
          "input": 12.0,
          "output": 36.0,
          "cache_hit": 2.4
        },
        "note": "CN card 2026-08-18: base input 12 / output 36 / cache-hit 2.4 per 1M. LIMITED-TIME 50% OFF active: input 6 / output 18 / cache 1.2 (card shows original+discounted). Parallel to qwen3.8-max night promo framing.",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 12.0,
        "output": 36.0,
        "cache_hit": 2.4,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18",
        "night/campaign_promo": {
          "kind": "limited-time 50% off",
          "note": "Card shows original and discounted rate; effective 6/18/1.2 per 1M once promo ends this resets to 12/36/2.4"
        }
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 262144,
        "reasoning_effort": {
          "validation_enum": [
            "none",
            "minimal",
            "low",
            "medium",
            "high",
            "xhigh"
          ],
          "enum_note": "Gateway validation acceptance enumerated 2026-08-05 via zero-cost invalid-value probe. Acceptance, not behaviour \u2014 the lists are partly a generic superset (2026-07-24 sweep); per-model differences here match July's behavioural findings."
        }
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen3.7-max",
        "cn": "https://www.qianwenai.com/models/qwen3.7-max"
      }
    },
    {
      "id": "qwen3.7-plus",
      "vendor": "Qwen",
      "kind": "chat",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-06-02",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 131072
      },
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": true,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": true,
        "vision_tool_result": true,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 39,
      "aa_url": "https://artificialanalysis.ai/models/qwen3-7-plus",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet (limited-time 8\u6298 promo active)",
        "per_1m_rmb": {
          "input": 2.0,
          "output": 8.0,
          "cache_hit": 0.4
        },
        "note": "CN card 2026-08-18: base input 2 / output 8 / cache-hit 0.4 per 1M. LIMITED-TIME 80% promo active: input 1.6 / output 6.4 / cache 0.32.",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 2.0,
        "output": 8.0,
        "cache_hit": 0.4,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18",
        "campaign_promo": {
          "kind": "limited-time 8\u6298",
          "note": "Effective 1.6/6.4/0.32 per 1M during promo; base 2/8/0.4 afterwards"
        }
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 262144,
        "reasoning_effort": {
          "validation_enum": [
            "none",
            "minimal",
            "low",
            "medium",
            "high",
            "xhigh"
          ],
          "enum_note": "Gateway validation acceptance enumerated 2026-08-05 via zero-cost invalid-value probe. Acceptance, not behaviour \u2014 the lists are partly a generic superset (2026-07-24 sweep); per-model differences here match July's behavioural findings."
        }
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen3.7-plus",
        "cn": "https://www.qianwenai.com/models/qwen3.7-plus"
      }
    },
    {
      "id": "qwen3.6-plus",
      "vendor": "Qwen",
      "kind": "chat",
      "tiers": [
        "team"
      ],
      "released": "2026-04-02",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 65536
      },
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": true,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": true,
        "vision_tool_result": true,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 40,
      "aa_url": "https://artificialanalysis.ai/models/qwen3-6-plus",
      "status_note": "AA marks deprecated (superseded by qwen3.7-plus) as of 2026-08-18; gateway still serves it.",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 2.0,
          "output": 12.0
        },
        "note": "CN card 2026-08-18: input 2 / output 12 per 1M. Explicit-cache create 2.5 / hit 0.2; batch-file 1/6 (Model Studio, not Token Plan route). No implicit cache-hit row shown on card.",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 2.0,
        "output": 12.0,
        "source": "qwenianwenai.com marketplace card (source qianwenai.com marketplace card)",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 131072
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen3.6-plus",
        "cn": "https://www.qianwenai.com/models/qwen3.6-plus"
      }
    },
    {
      "id": "qwen3.6-flash",
      "vendor": "Qwen",
      "kind": "chat",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-04-27",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 65536
      },
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": true,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": true,
        "vision_tool_result": true,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 32,
      "aa_url": "https://artificialanalysis.ai/models/qwen3-6-35b-a3b",
      "status_note": "Token Plan id IS the open-weights qwen3.6-35b-a3b; AA tracks the 35b-a3b weights identity",
      "aliases": [
        "qwen3.6-35b-a3b",
        "qwen3.6-flash-2026-04-16"
      ],
      "underlying_identity": "qwen3.6-35b-a3b",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 1.2,
          "output": 7.2
        },
        "note": "CN card 2026-08-18: input 1.2 / output 7.2 per 1M. Explicit-cache create 1.5 / hit 0.12; batch-file 0.6/3.6. No implicit cache-hit row shown.",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 1.2,
        "output": 7.2,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 131072,
        "reasoning_effort": {
          "validation_enum": [
            "none",
            "minimal",
            "low",
            "medium",
            "high",
            "xhigh"
          ],
          "enum_note": "Gateway validation acceptance enumerated 2026-08-05 via zero-cost invalid-value probe. Acceptance, not behaviour \u2014 the lists are partly a generic superset (2026-07-24 sweep); per-model differences here match July's behavioural findings."
        }
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen3.6-flash",
        "cn": "https://www.qianwenai.com/models/qwen3.6-flash"
      }
    },
    {
      "id": "deepseek-v4-pro",
      "vendor": "DeepSeek",
      "kind": "chat",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-04-24",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 384000
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": false,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": false
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 53,
      "aa_url": "https://artificialanalysis.ai/models/deepseek-v4-pro",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 12.0,
          "output": 24.0,
          "cache_hit": 1.0
        },
        "note": "CN card 2026-08-18: input 12 / output 24 / cache-hit 1 per 1M. Base model (the -0813 dated variant has its own lower rate; record them separately).",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 12.0,
        "output": 24.0,
        "cache_hit": 1.0,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 393216,
        "reasoning_effort": {
          "validation_enum": [
            "low",
            "medium",
            "high",
            "xhigh",
            "max"
          ],
          "enum_note": "Gateway validation acceptance enumerated 2026-08-05 via zero-cost invalid-value probe. Acceptance, not behaviour \u2014 the lists are partly a generic superset (2026-07-24 sweep); per-model differences here match July's behavioural findings."
        }
      },
      "card": {
        "en": "https://www.qwencloud.com/models/deepseek-v4-pro",
        "cn": "https://www.qianwenai.com/models/deepseek-v4-pro"
      }
    },
    {
      "id": "deepseek-v4-pro-0813",
      "vendor": "DeepSeek",
      "kind": "chat",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-08-13",
      "status": "unlisted",
      "limit": {
        "context": 1000000,
        "output": 384000
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": false,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": false
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 53,
      "aa_url": "https://artificialanalysis.ai/models/deepseek-v4-pro",
      "status_note": "Servable by exact id (HTTP 200, probed 2026-08-18) but deliberately NOT surfaced on the gateway /models listing. Unlike 'retired' it must stay in downstream catalogues (it is callable); unlike 'active' it is excluded from the check-drift expected /models set (else it would false-alarm as removed every run). Treat /models gain as real drift when it eventually lists. | deepseek-v4-pro-0813 is the dated 2026-08-13 release; AA tracks it as the current deepseek-v4-pro id",
      "aliases": [
        "deepseek-v4-pro"
      ],
      "underlying_identity": "deepseek-v4-pro",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 4.5,
          "output": 13.5,
          "cache_hit": 0.45
        },
        "per_1m_rmb_tier2": {
          "input": 9,
          "output": 27,
          "cache_hit": 0.9
        },
        "note": "CN qianwenai.com marketplace card 2026-08-18: two pricing bands per the tiered input-band scheme. Tier 1 input 4.5 / output 13.5 / cache-hit 0.45 per 1M; tier 2 input 9 / output 27 / cache-hit 0.9 (exactly 2x tier 1). This is the dated -0813 variant's OWN card rate - it does NOT inherit the base deepseek-v4-pro 12/24 rate (that base figure is separately recorded). Card also shows batch-task and fine-tune flags; those describe Model Studio platform services, NOT Token Plan gateway routes (batch/fine-tuning 404 on the Token Plan gateway, proven 2026-08-04). LIMITED-TIME NIGHT PROMO (both Team & Personal, EN/CN overviews): 50% off Credits between 22:00-08:00 UTC+8, parallel to the qwen3.8-max night discount.",
        "night_promo": {
          "discount": "50% off Credits",
          "window": "22:00 - 08:00 UTC+8",
          "note": "Confirmed on EN/CN Team + Personal overview pages since 2026-08-15 doc snapshot."
        },
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 4.5,
        "output": 13.5,
        "cache_hit": 0.45,
        "source": "qianwenai.com marketplace card (deepseek-v4-pro-0813)",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 393216,
        "reasoning_effort": {
          "validation_enum": [
            "low",
            "medium",
            "high",
            "xhigh",
            "max"
          ],
          "enum_note": "Inherited from base deepseek-v4-pro (2026-08-05 enumeration); the dated variant was not re-enumerated. Same family, acceptance assumed until probed."
        }
      },
      "card": {
        "en": "https://www.qwencloud.com/models/deepseek-v4-pro-0813",
        "cn": "https://www.qianwenai.com/models/deepseek-v4-pro-0813"
      }
    },
    {
      "id": "deepseek-v4-flash",
      "vendor": "DeepSeek",
      "kind": "chat",
      "tiers": [
        "team"
      ],
      "released": "2026-04-24",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 384000
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": false,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": false
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 52,
      "aa_url": "https://artificialanalysis.ai/models/deepseek-v4-flash",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 1.0,
          "output": 2.0,
          "cache_hit": 0.2
        },
        "note": "CN card 2026-08-18: input 1 / output 2 / cache-hit 0.2 per 1M.",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 1.0,
        "output": 2.0,
        "cache_hit": 0.2,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 393216
      },
      "card": {
        "en": "https://www.qwencloud.com/models/deepseek-v4-flash",
        "cn": "https://www.qianwenai.com/models/deepseek-v4-flash"
      }
    },
    {
      "id": "deepseek-v4-flash-0731",
      "vendor": "DeepSeek",
      "kind": "chat",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-08-01",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 393216
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "json_schema",
        "responses_api": true,
        "logprobs": false,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": false
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 52,
      "aa_url": "https://artificialanalysis.ai/models/deepseek-v4-flash",
      "status_note": "deepseek-v4-flash-0731 is the dated 2026-08-01 release; AA tracks it as the current deepseek-v4-flash id",
      "aliases": [
        "deepseek-v4-flash"
      ],
      "underlying_identity": "deepseek-v4-flash",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 1,
          "output": 2,
          "cache_hit": 0.2
        },
        "note": "Marketplace card 2026-08-05: input \u00a51/M, output \u00a52/M, cache-hit \u00a50.2/M (=100/200/20 credits per 1M). Card also shows batch-task and fine-tune flags \u2014 those describe Model Studio platform services, NOT Token Plan gateway routes (batch/fine-tuning routes 404 on the Token Plan gateway, proven 2026-08-04).",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 1,
        "output": 2,
        "cache_hit": 0.2,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-05"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "reasoning_effort": {
          "validation_enum": [
            "low",
            "medium",
            "high",
            "xhigh",
            "max"
          ],
          "enum_note": "Gateway validation acceptance enumerated 2026-08-05 via zero-cost invalid-value probe. Acceptance, not behaviour \u2014 the lists are partly a generic superset (2026-07-24 sweep); per-model differences here match July's behavioural findings."
        },
        "budget_max": 393216,
        "note": "First thinking records from the 2026-08-05 free enumeration: budget ceiling 393,216 (matches deepseek-v4-pro) and deepseek's characteristic no-none/no-minimal validation enum."
      },
      "card": {
        "en": "https://www.qwencloud.com/models/deepseek-v4-flash-0731",
        "cn": "https://www.qianwenai.com/models/deepseek-v4-flash-0731"
      }
    },
    {
      "id": "deepseek-v3.2",
      "vendor": "DeepSeek",
      "kind": "chat",
      "tiers": [
        "team"
      ],
      "released": "2025-12-03",
      "status": "active",
      "limit": {
        "context": 131072,
        "output": 65536
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "json_schema",
        "responses_api": false,
        "logprobs": false,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": false
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 25,
      "aa_url": "https://artificialanalysis.ai/models/deepseek-v3-2",
      "status_note": "AA marks deprecated (superseded by deepseek-v4 pro family) as of 2026-08-18; gateway still serves it.",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 2.0,
          "output": 3.0,
          "cache_hit": 0.4
        },
        "note": "CN card 2026-08-18: input 2 / output 3 / cache-hit 0.4 per 1M. Batch-file 1/1.5 per 1M exists on card (Model Studio, not Token Plan route).",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 2.0,
        "output": 3.0,
        "cache_hit": 0.4,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": null
      },
      "retires": "2026-10-10",
      "card": {
        "en": "https://www.qwencloud.com/models/deepseek-v3.2",
        "cn": "https://www.qianwenai.com/models/deepseek-v3.2"
      }
    },
    {
      "id": "kimi-k2.7-code",
      "vendor": "Moonshot",
      "kind": "chat",
      "tiers": [
        "team"
      ],
      "released": "2026-06-12",
      "status": "active",
      "limit": {
        "context": 262144,
        "output": 262144
      },
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": true,
        "structured_output": "json_schema",
        "responses_api": false,
        "logprobs": true,
        "vision_tool_result": true,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 43,
      "aa_url": "https://artificialanalysis.ai/models/kimi-k2-7-code",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 6.5,
          "output": 27.0,
          "cache_hit": 1.3
        },
        "note": "CN card 2026-08-18: input 6.5 / output 27 / cache-hit 1.3 per 1M; explicit-cache create 8.125 / hit 0.65.",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 6.5,
        "output": 27.0,
        "cache_hit": 1.3,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 262144
      },
      "card": {
        "en": "https://www.qwencloud.com/models/kimi-k2.7-code",
        "cn": "https://www.qianwenai.com/models/kimi-k2.7-code"
      }
    },
    {
      "id": "kimi-k2.6",
      "vendor": "Moonshot",
      "kind": "chat",
      "tiers": [
        "team"
      ],
      "released": "2026-04-21",
      "status": "active",
      "limit": {
        "context": 262144,
        "output": 262144
      },
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": true,
        "structured_output": "json_schema",
        "responses_api": false,
        "logprobs": true,
        "vision_tool_result": true,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 45,
      "aa_url": "https://artificialanalysis.ai/models/kimi-k2-6",
      "status_note": "AA marks deprecated (superseded by kimi-k3) as of 2026-08-18; gateway still serves it.",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 6.5,
          "output": 27.0,
          "cache_hit": 1.3
        },
        "note": "CN card 2026-08-18: input 6.5 / output 27 / cache-hit 1.3 per 1M; explicit-cache create 8.125 / hit 0.65. CN-only (no EN card).",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 6.5,
        "output": 27.0,
        "cache_hit": 1.3,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": null
      },
      "card": {
        "cn": "https://www.qianwenai.com/models/kimi-k2.6"
      }
    },
    {
      "id": "kimi-k2.5",
      "vendor": "Moonshot",
      "kind": "chat",
      "tiers": [
        "team"
      ],
      "released": "2026-01-01",
      "status": "active",
      "limit": {
        "context": 262144,
        "output": 98304
      },
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": true,
        "structured_output": "json_schema",
        "responses_api": false,
        "logprobs": true,
        "vision_tool_result": true,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 36,
      "aa_url": "https://artificialanalysis.ai/models/kimi-k2-5",
      "status_note": "AA marks deprecated (superseded by kimi-k2.6) as of 2026-08-18; gateway still serves it.",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 4.0,
          "output": 21.0,
          "cache_hit": 0.8
        },
        "note": "CN card 2026-08-18: input 4 / output 21 / cache-hit 0.8 per 1M; explicit-cache create 5 / hit 0.4. CN-only (no EN card).",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 4.0,
        "output": 21.0,
        "cache_hit": 0.8,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": null
      },
      "card": {
        "cn": "https://www.qianwenai.com/models/kimi-k2.5"
      }
    },
    {
      "id": "glm-5.2",
      "vendor": "Zhipu",
      "kind": "chat",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-06-13",
      "status": "active",
      "limit": {
        "context": 1000000,
        "output": 131072
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "json_schema",
        "responses_api": false,
        "logprobs": false,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 53,
      "aa_url": "https://artificialanalysis.ai/models/glm-5-2",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 8.0,
          "output": 28.0,
          "cache_hit": 2.0
        },
        "note": "CN card 2026-08-18: input 8 / output 28 / cache-hit 2 per 1M.",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 8.0,
        "output": 28.0,
        "cache_hit": 2.0,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "reasoning_effort": {
          "accepted": [
            "high",
            "max"
          ],
          "validation_enum": [
            "none",
            "minimal",
            "low",
            "medium",
            "high",
            "xhigh",
            "max"
          ],
          "enum_note": "Gateway validation acceptance enumerated 2026-08-05 via zero-cost invalid-value probe. Acceptance, not behaviour \u2014 the lists are partly a generic superset (2026-07-24 sweep); per-model differences here match July's behavioural findings."
        },
        "budget_max": 131072
      },
      "card": {
        "en": "https://www.qwencloud.com/models/glm-5.2",
        "cn": "https://www.qianwenai.com/models/glm-5.2"
      }
    },
    {
      "id": "glm-5.1",
      "vendor": "Zhipu",
      "kind": "chat",
      "tiers": [
        "team"
      ],
      "released": "2026-04-07",
      "status": "active",
      "limit": {
        "context": 202752,
        "output": 128000
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "json_schema",
        "responses_api": false,
        "logprobs": true,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": false
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 41,
      "aa_url": "https://artificialanalysis.ai/models/glm-5-1",
      "status_note": "AA marks deprecated (superseded by glm-5.2) as of 2026-08-18; gateway still serves it.",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 6.0,
          "output": 24.0,
          "cache_hit": 1.2
        },
        "note": "CN card 2026-08-18: input 6 / output 24 / cache-hit 1.2 per 1M; explicit-cache create 7.5 / hit 0.6.",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 6.0,
        "output": 24.0,
        "cache_hit": 1.2,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 131072
      },
      "card": {
        "en": "https://www.qwencloud.com/models/glm-5.1",
        "cn": "https://www.qianwenai.com/models/glm-5.1"
      }
    },
    {
      "id": "glm-5",
      "vendor": "Zhipu",
      "kind": "chat",
      "tiers": [
        "team"
      ],
      "released": "2026-02-12",
      "status": "active",
      "limit": {
        "context": 202752,
        "output": 16384
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "json_schema",
        "responses_api": false,
        "logprobs": true,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": false
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 41,
      "aa_url": "https://artificialanalysis.ai/models/glm-5",
      "status_note": "AA marks deprecated (superseded by glm-5.1) as of 2026-08-18; gateway still serves it.",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 4.0,
          "output": 18.0,
          "cache_hit": 0.8
        },
        "note": "CN card 2026-08-18: input 4 / output 18 / cache-hit 0.8 per 1M. CN-only (no EN card).",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 4.0,
        "output": 18.0,
        "cache_hit": 0.8,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "hybrid",
        "can_disable": true,
        "budget_max": 131072
      },
      "card": {
        "cn": "https://www.qianwenai.com/models/glm-5"
      }
    },
    {
      "id": "MiniMax-M2.5",
      "vendor": "MiniMax",
      "kind": "chat",
      "tiers": [
        "team"
      ],
      "released": "2026-02-12",
      "status": "active",
      "limit": {
        "context": 196608,
        "output": 32768
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "capabilities": {
        "reasoning": true,
        "tool_call": true,
        "attachment": false,
        "structured_output": "none",
        "responses_api": false,
        "logprobs": false,
        "vision_tool_result": false,
        "prompt_cache_key_accepted": true,
        "max_tokens_range_validated": true
      },
      "endpoint": "/compatible-mode/v1/chat/completions",
      "aa_intelligence": 34,
      "aa_url": "https://artificialanalysis.ai/models/minimax-m2-5",
      "status_note": "AA marks deprecated (superseded by MiniMax-M2.7) as of 2026-08-18; gateway still serves it.",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet",
        "per_1m_rmb": {
          "input": 2.1,
          "output": 8.4,
          "cache_hit": 0.42
        },
        "note": "CN card 2026-08-18: input 2.1 / output 8.4 / cache-hit 0.42 per 1M. CN-only (no EN card).",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 2.1,
        "output": 8.4,
        "cache_hit": 0.42,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18"
      },
      "thinking": {
        "mode": "always_on",
        "can_disable": false,
        "budget_max": 38912
      },
      "card": {
        "cn": "https://www.qianwenai.com/models/MiniMax-M2.5"
      }
    },
    {
      "id": "qwen-audio-3.0-tts-plus",
      "vendor": "Qwen",
      "kind": "tts",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-07-14",
      "status": "active",
      "limit": {
        "context": null,
        "output": null,
        "note": "Streaming TTS, character-metered (no token context window). RPM 180."
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "audio"
        ]
      },
      "capabilities": {
        "reasoning": false,
        "tool_call": false,
        "attachment": false,
        "structured_output": "none",
        "responses_api": false,
        "logprobs": false,
        "note": "documented/unprobed (2026-08-13): describes the OpenAI-compatible surface, where this model is NOT servable at all \u2014 every flag is false by construction. The DashScope realtime surface has its own contract, unprobed under the spend gate.",
        "task_runs_on_gateway": false
      },
      "endpoint": "wss://token-plan.<region>.maas.aliyuncs.com/api-ws/v1/inference (DashScope WebSocket protocol; set dashscope.base_websocket_api_url per the vendor's Token Plan multimodal best-practices page). VERIFIED 2026-08-14 by handshake probe (probe-plan-routes.py --ws): HTTP 101 on both regions, auth enforced at handshake (bad key -> 401 InvalidApiKey), zero token cost. Task-level probe run 2026-08-14: the handshake and run-task are accepted (task-started) and the engine then fails with '[cosyvoice:] Engine error [411]', for every voice tried including an invalid one, so the model is addressable but NOT runnable.",
      "status_note": "Listed on /models both tiers, but synthesis is broken vendor-side as of 2026-08-14: every key (Personal, Team, Team-with-quota) gets '[cosyvoice:]Engine error [411]' after task-started on the documented WebSocket surface, reproduced with the official SDK. Advertised but not currently servable.",
      "cost": {
        "status": "published, character-metered",
        "per_10k_chars_rmb": 1.4,
        "note": "Marketplace card 2026-08-05: \u00a51.4 per \u4e07\u5b57\u7b26 (10k characters), RPM 180. Character-metered, not token-metered.",
        "published": true,
        "currency": "RMB",
        "unit": "per_10k_chars",
        "input": 1.4,
        "output": null,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-05"
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen-audio-3.0-tts-plus",
        "cn": "https://www.qianwenai.com/models/qwen-audio-3.0-tts-plus"
      }
    },
    {
      "id": "qwen-audio-3.0-realtime-plus",
      "vendor": "Qwen",
      "kind": "realtime",
      "tiers": [
        "team",
        "personal"
      ],
      "status": "active",
      "limit": {
        "context": null,
        "output": null,
        "note": "Realtime streaming audio dialog (no fixed token context window). Per-token priced."
      },
      "modalities": {
        "input": [
          "audio",
          "text"
        ],
        "output": [
          "audio",
          "text"
        ]
      },
      "capabilities": {
        "reasoning": false,
        "tool_call": false,
        "attachment": true,
        "structured_output": "none",
        "responses_api": false,
        "logprobs": false,
        "note": "documented/unprobed (2026-08-13): describes the OpenAI-compatible surface, where this model is NOT servable at all \u2014 every flag is false by construction. The DashScope realtime surface has its own contract, unprobed under the spend gate.",
        "task_runs_on_gateway": false
      },
      "endpoint": "wss://token-plan.<region>.maas.aliyuncs.com/api-ws/v1/inference (DashScope WebSocket protocol; set dashscope.base_websocket_api_url per the vendor's Token Plan multimodal best-practices page). VERIFIED 2026-08-14 by handshake probe (probe-plan-routes.py --ws): HTTP 101 on both regions, auth enforced at handshake (bad key -> 401 InvalidApiKey), zero token cost. Task-level probe run 2026-08-14: this model is not served on the inference path at all ('url error, please check url'); the realtime routes /api-ws/v1/realtime and /api-ws/v1/realtime-inference exist and handshake, but the protocol is client-initiated and a session was not attempted. Realtime routes discovered 2026-08-14: wss://token-plan.<region>.maas.aliyuncs.com/api-ws/v1/realtime and /api-ws/v1/realtime-inference (101 + auth; the generic /api-ws/v1/inference path rejects this model with 'url error').",
      "cost": {
        "status": "published per-token pricing, CN-denominated sheet (realtime audio)",
        "per_1m_rmb": {
          "input": 5.0,
          "output": 40.0
        },
        "note": "Realtime voice card 2026-08-18: TEXT input 5 / TEXT output 40; AUDIO input 40; text+audio output 150 per 1M (output text not billed when audio present). surface is DashScope webSocket realtime, not the compatible-mode chat route.",
        "published": true,
        "currency": "RMB",
        "unit": "per_1m_tokens",
        "input": 5.0,
        "output": 40.0,
        "source": "qianwenai.com marketplace card",
        "verified": "2026-08-18",
        "audio_input_per_1m": 40,
        "audio_output_textplusaudio_per_1m": 150
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen-audio-3.0-realtime-plus",
        "cn": "https://www.qianwenai.com/models/qwen-audio-3.0-realtime-plus"
      }
    },
    {
      "id": "qwen-image-2.0",
      "vendor": "Qwen",
      "kind": "image",
      "tiers": [
        "team"
      ],
      "released": "2026-03-03",
      "status": "active",
      "limit": {
        "context": 8192,
        "output": null
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "image"
        ]
      },
      "capabilities": {
        "reasoning": false,
        "tool_call": false,
        "attachment": false,
        "structured_output": "none"
      },
      "endpoint": "unverified \u2014 listed on /models but the generation endpoint has not been probed",
      "cost": {
        "status": "published, per-image",
        "source": "qianwenai.com + qwencloud.com marketplace cards",
        "verified": "2026-08-18",
        "currency": "RMB",
        "unit": "per_image",
        "input": 0.2,
        "output": null,
        "limits": {
          "rpm": 120
        },
        "note": "\u56fe\u7247\u751f\u6210 \u00a50.2/\u5f20 (per image); RPM 120."
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen-image-2.0",
        "cn": "https://www.qianwenai.com/models/qwen-image-2.0"
      }
    },
    {
      "id": "qwen-image-2.0-pro",
      "vendor": "Qwen",
      "kind": "image",
      "tiers": [
        "team"
      ],
      "released": "2026-03-03",
      "status": "active",
      "limit": {
        "context": 8192,
        "output": null
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "image"
        ]
      },
      "capabilities": {
        "reasoning": false,
        "tool_call": false,
        "attachment": false,
        "structured_output": "none"
      },
      "endpoint": "unverified \u2014 listed on /models but the generation endpoint has not been probed",
      "cost": {
        "status": "published, per-image",
        "source": "qianwenai.com + qwencloud.com marketplace cards",
        "verified": "2026-08-18",
        "currency": "RMB",
        "unit": "per_image",
        "input": 0.5,
        "output": null,
        "limits": {
          "rpm": 2
        },
        "note": "\u56fe\u7247\u751f\u6210 \u00a50.5/\u5f20 (per image); RPM 2."
      },
      "card": {
        "en": "https://www.qwencloud.com/models/qwen-image-2.0-pro",
        "cn": "https://www.qianwenai.com/models/qwen-image-2.0-pro"
      }
    },
    {
      "id": "wan2.7-image",
      "vendor": "Wan",
      "kind": "image",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-05-29",
      "status": "active",
      "limit": {
        "context": 8192,
        "output": null
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "image"
        ]
      },
      "capabilities": {
        "reasoning": false,
        "tool_call": false,
        "attachment": false,
        "structured_output": "none"
      },
      "endpoint": "unverified \u2014 listed on /models but the generation endpoint has not been probed",
      "cost": {
        "status": "published, per-image",
        "source": "qianwenai.com + qwencloud.com marketplace cards",
        "verified": "2026-08-18",
        "currency": "RMB",
        "unit": "per_image",
        "input": 0.2,
        "output": null,
        "limits": {
          "rpm": 300,
          "concurrency": 5
        },
        "note": "\u56fe\u7247\u751f\u6210 \u00a50.2/\u5f20 (per image); RPM 300."
      },
      "card": {
        "en": "https://www.qwencloud.com/models/wan2.7-image",
        "cn": "https://www.qianwenai.com/models/wan2.7-image"
      }
    },
    {
      "id": "wan2.7-image-pro",
      "vendor": "Wan",
      "kind": "image",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-05-29",
      "status": "active",
      "limit": {
        "context": 8192,
        "output": null
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "image"
        ]
      },
      "capabilities": {
        "reasoning": false,
        "tool_call": false,
        "attachment": false,
        "structured_output": "none"
      },
      "endpoint": "unverified \u2014 listed on /models but the generation endpoint has not been probed",
      "cost": {
        "status": "published, per-image",
        "source": "qianwenai.com + qwencloud.com marketplace cards",
        "verified": "2026-08-18",
        "currency": "RMB",
        "unit": "per_image",
        "input": 0.5,
        "output": null,
        "limits": {
          "rpm": 300,
          "concurrency": 5
        },
        "note": "\u56fe\u7247\u751f\u6210 \u00a50.5/\u5f20 (per image); RPM 300."
      },
      "card": {
        "en": "https://www.qwencloud.com/models/wan2.7-image-pro",
        "cn": "https://www.qianwenai.com/models/wan2.7-image-pro"
      }
    },
    {
      "id": "happyhorse-1.1-t2v",
      "vendor": "HappyHorse",
      "kind": "video",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-07-17",
      "status": "active",
      "limit": {
        "context": null,
        "output": null
      },
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "video"
        ]
      },
      "capabilities": {
        "reasoning": false,
        "tool_call": false,
        "attachment": false,
        "structured_output": "none"
      },
      "endpoint": "/api/v1/services/aigc/video-generation/video-synthesis (async, X-DashScope-Async: enable, poll /api/v1/tasks/{task_id})",
      "cost": {
        "status": "published, per-second video (tiered by resolution)",
        "source": "qianwenai.com + qwencloud.com marketplace cards",
        "verified": "2026-08-18",
        "currency": "RMB",
        "unit": "per_second_video",
        "rates": {
          "480p": {
            "list": 0.45,
            "promo": 0.27,
            "promo_note": "\u9650\u65f66\u6298 (CN)"
          },
          "720p": {
            "list": 0.9,
            "promo": 0.54,
            "promo_note": "\u9650\u65f66\u6298 (CN)"
          },
          "1080p": {
            "list": 1.2,
            "promo": 0.72,
            "promo_note": "\u9650\u65f66\u6298 (CN)"
          }
        },
        "limits": {
          "concurrency": 5,
          "async_queue": 500,
          "rpm": 300
        },
        "note": "CN card: \u89c6\u9891\u751f\u6210 480P \u00a50.45/s(6\u6298\u00a50.27), 720P \u00a50.9/s(6\u6298\u00a50.54), 1080P \u00a51.2/s(6\u6298\u00a50.72). EN card: 480P $0.07\u21920.042/s, 720P $0.14\u21920.084/s, 1080P $0.18\u21920.108/s (40% off). Same rate structure across i2v/r2v/t2v."
      },
      "card": {
        "en": "https://www.qwencloud.com/models/happyhorse-1.1-t2v",
        "cn": "https://www.qianwenai.com/models/happyhorse-1.1-t2v"
      }
    },
    {
      "id": "happyhorse-1.1-i2v",
      "vendor": "HappyHorse",
      "kind": "video",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-07-17",
      "status": "active",
      "limit": {
        "context": null,
        "output": null
      },
      "modalities": {
        "input": [
          "image",
          "text"
        ],
        "output": [
          "video"
        ]
      },
      "capabilities": {
        "reasoning": false,
        "tool_call": false,
        "attachment": true,
        "structured_output": "none"
      },
      "endpoint": "/api/v1/services/aigc/video-generation/video-synthesis (async, X-DashScope-Async: enable, poll /api/v1/tasks/{task_id})",
      "cost": {
        "status": "published, per-second video (tiered by resolution)",
        "source": "qianwenai.com + qwencloud.com marketplace cards",
        "verified": "2026-08-18",
        "currency": "RMB",
        "unit": "per_second_video",
        "rates": {
          "480p": {
            "list": 0.45,
            "promo": 0.27,
            "promo_note": "\u9650\u65f66\u6298 (CN)"
          },
          "720p": {
            "list": 0.9,
            "promo": 0.54,
            "promo_note": "\u9650\u65f66\u6298 (CN)"
          },
          "1080p": {
            "list": 1.2,
            "promo": 0.72,
            "promo_note": "\u9650\u65f66\u6298 (CN)"
          }
        },
        "limits": {
          "concurrency": 5,
          "async_queue": 500,
          "rpm": 300
        },
        "note": "CN card: \u89c6\u9891\u751f\u6210 480P \u00a50.45/s(6\u6298\u00a50.27), 720P \u00a50.9/s(6\u6298\u00a50.54), 1080P \u00a51.2/s(6\u6298\u00a50.72). EN card: 480P $0.07\u21920.042/s, 720P $0.14\u21920.084/s, 1080P $0.18\u21920.108/s (40% off). Same rate structure across i2v/r2v/t2v."
      },
      "card": {
        "en": "https://www.qwencloud.com/models/happyhorse-1.1-i2v",
        "cn": "https://www.qianwenai.com/models/happyhorse-1.1-i2v"
      }
    },
    {
      "id": "happyhorse-1.1-r2v",
      "vendor": "HappyHorse",
      "kind": "video",
      "tiers": [
        "team",
        "personal"
      ],
      "released": "2026-07-17",
      "status": "active",
      "limit": {
        "context": null,
        "output": null
      },
      "modalities": {
        "input": [
          "image",
          "text"
        ],
        "output": [
          "video"
        ]
      },
      "capabilities": {
        "reasoning": false,
        "tool_call": false,
        "attachment": true,
        "structured_output": "none"
      },
      "endpoint": "/api/v1/services/aigc/video-generation/video-synthesis (async, X-DashScope-Async: enable, poll /api/v1/tasks/{task_id})",
      "cost": {
        "status": "published, per-second video (tiered by resolution)",
        "source": "qianwenai.com + qwencloud.com marketplace cards",
        "verified": "2026-08-18",
        "currency": "RMB",
        "unit": "per_second_video",
        "rates": {
          "480p": {
            "list": 0.45,
            "promo": 0.27,
            "promo_note": "\u9650\u65f66\u6298 (CN)"
          },
          "720p": {
            "list": 0.9,
            "promo": 0.54,
            "promo_note": "\u9650\u65f66\u6298 (CN)"
          },
          "1080p": {
            "list": 1.2,
            "promo": 0.72,
            "promo_note": "\u9650\u65f66\u6298 (CN)"
          }
        },
        "limits": {
          "concurrency": 5,
          "async_queue": 500,
          "rpm": 300
        },
        "note": "CN card: \u89c6\u9891\u751f\u6210 480P \u00a50.45/s(6\u6298\u00a50.27), 720P \u00a50.9/s(6\u6298\u00a50.54), 1080P \u00a51.2/s(6\u6298\u00a50.72). EN card: 480P $0.07\u21920.042/s, 720P $0.14\u21920.084/s, 1080P $0.18\u21920.108/s (40% off). Same rate structure across i2v/r2v/t2v."
      },
      "card": {
        "en": "https://www.qwencloud.com/models/happyhorse-1.1-r2v",
        "cn": "https://www.qianwenai.com/models/happyhorse-1.1-r2v"
      }
    }
  ]
}