{"auto_groups":["Seedance_1","Seedance","default"],"data":[{"model_name":"gpt-5.6-sol","description":"GPT-5.6 Sol是该系列旗舰，面向前沿推理与长程 Agent 任务，在编码、知识工作、网络安全与科学研究等场景达到业界领先水平，以更少 token 与更高性价比完成复杂工作","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":6,"enable_groups":["AzureGPT","default","GPT","GPT_K","OpenAI","wy_vip"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 272000 ? tier(\"short context\", p * 4 + c * 20 + cr * 0.4 + cc * 5) : p \u003e 272000 ? tier(\"long context\", p * 8 + c * 30 + cr * 0.8 + cc * 10) : tier(\"default\", p * 8 + c * 30 + cr * 0.8 + cc * 10)","pricing_version":"5a90f2b86c08bd983a9a2e6d66c255f4eaef9c4bc934386d2b6ae84ef0ff1f1f"},{"model_name":"qwen3.6-plus","description":"Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。","icon":"Qwen.Color","tags":"对话,思考,联网搜索","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","default","Qwen","vip"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 256000 ? tier(\"short context\", p * 0.2941 + c * 1.7647 + cr * 0.0294) : p \u003e 256000 \u0026\u0026 p \u003c= 1000000 ? tier(\"long context\", p * 1.1765 + c * 7.0588 + cr * 0.1176) : tier(\"default\", p * 1.1765 + c * 7.0588 + cr * 0.1176)"},{"model_name":"gpt-6-sol","description":"GPT-6 Astra 是 OpenAI 面向复杂端到端任务打造的旗舰级模型，擅长高级推理、软件工程、计算机操作、研究分析、文档创建和多步骤工作流。该模型支持文本与图像输入，具备高性能 Agent 能力，适用于企业自动化和复杂生产任务。","icon":"OpenAI","tags":"对话,思考,联网思考","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":2,"enable_groups":["GPT","GPT_K","OpenAI","wy_vip","default"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 272000 ? tier(\"short context\", p * 2 + c * 10 + cr * 0.2 + cc * 2.5) : p \u003e 272000 ? tier(\"long context\", p * 4 + c * 15 + cr * 0.4 + cc * 5) : tier(\"default\", p * 4 + c * 15 + cr * 0.4 + cc * 5)"},{"model_name":"claude-haiku-4-5-20251001","description":"Claude Haiku 4.5 是 Anthropic 效率最高的模型，能够快速且经济高效地提供接近前沿的性能。具体来说，Claude Haiku 4.5 在编程、计算机使用和代理任务上的表现与 Claude Sonnet 4 相当，非常适合用于支持扩展的子代理部署、预算受限的应用和免费产品。此外，Claude Haiku 4.5 的速度使得使用计算机的代理更快、更响应迅速，对于响应时间至关重要的用例（如客服代理和聊天机器人）非常有利。\n输入：文本、图像\n输出：文本","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":0.5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["Anthropic_U","bingyu_vip","CCAnthropic","default","Anthropic"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"gpt-5.5","description":"GPT-5.5是OpenAI最先进的自主型AI模型，旨在在极少引导下独立规划并完成复杂、多步骤的任务——如编程、研究及数据分析——同时提升效率与整体智能水平。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":6,"enable_groups":["AzureGPT","bingyu_vip","default","GPT","GPT_K","OpenAI","unal_vip","wy_vip","xy_vip"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 272000 ? tier(\"short context\", p * 5 + c * 30 + cr * 0.5) : p \u003e 272000 ? tier(\"long context\", p * 10 + c * 45 + cr * 1) : tier(\"default\", p * 10 + c * 45 + cr * 1)"},{"model_name":"gemini-3-pro-image","icon":"Gemini.Color","tags":"文生图,图生图","vendor_id":3,"quota_type":0,"model_ratio":1,"model_price":0,"owner_by":"","completion_ratio":6,"cache_ratio":0.1,"image_ratio":1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai","image-generation"]},{"model_name":"gemini-3.5-flash-lite","icon":"Gemini.Color","quota_type":0,"model_ratio":0.15,"model_price":0,"owner_by":"","completion_ratio":8.333333333333,"cache_ratio":0.1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"claude-sonnet-4-6","description":"Claude Sonnet 4.6 是Anthropic迄今为止最强大的Sonnet模型。它全面提升了模型在编码、计算机操作、长上下文推理、代理规划、知识工作和设计等领域的技能。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":1.5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["Anthropic","Anthropic_U","AwsAnthropic","bingyu_vip","CCAnthropic","default"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"qwen3.6-flash","description":"阿里通义 Qwen3.6 系列中的轻量高效模型，原生支持文本、图像、视频多模态输入，上下文窗口达 100 万 token，支持工具调用与函数调用。相比上一代 3.5-Flash，在代码智能体、数学推理及视觉空间定位（物体检测/定位）方面均有明显提升，价格低廉，适合大规模高并发场景。","icon":"Qwen.Color","tags":"对话,思考,联网搜索","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","Qwen","bingyu_vip"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 256000 ? tier(\"short context\", p * 0.1765 + c * 1.0588 + cr * 0.0176) : p \u003e 256000 \u0026\u0026 p \u003c 1000000 ? tier(\"long context\", p * 0.7059 + c * 4.2353 + cr * 0.0176) : tier(\"default\", p * 0.7059 + c * 4.2353 + cr * 0.0176)"},{"model_name":"gpt-5.1-codex-mini","description":"GPT-5.1-codex-mini：轻量高效，专为资源受限、成本敏感场景打造（如教育、初创企业），性能接近主力模型，依然保留多模态和安全能力。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.125,"model_price":0,"owner_by":"","completion_ratio":8,"cache_ratio":0.12,"enable_groups":["default","OpenAI","unal_vip","xy_vip"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"gpt-5.2","description":"GPT-5.2 在 GPT-5.1 的基础上进行了改进，使其交互更加可靠、灵活和用户友好。重点在于提供更清晰的响应、更好的适应性和增强的多任务控制能力。主要亮点包括：\n1）加速代理开发：为分析、决策支持、代码现代化和数据管道构建 AI 代理。\n2）更高的连贯性：生成更符合用户指令的、可预测且准确的输出。\n自适应推理：处理复杂查询时能更深入，同时在简单任务上保持高效。\n3）扩展的上下文处理：支持更长的输入，适用于多步骤工作流程和更丰富的对话。\n4）精细化的定制：提供更好的对语气、风格和结构化输出的控制，以实现个性化体验。\n5）更广泛的应用性：在保持清晰度和安全性的同时，提升多个领域的性能。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.875,"model_price":0,"owner_by":"","completion_ratio":8,"cache_ratio":0.1,"enable_groups":["default","OpenAI","unal_vip","xy_vip"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"gpt-5.6-terra","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":6,"enable_groups":["default","GPT","GPT_K","OpenAI","wy_vip"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 272000 ? tier(\"short context\", p * 2 + c * 12 + cr * 0.2 + cc * 2.5) : p \u003e 272000 ? tier(\"long context\", p * 4 + c * 18 + cr * 0.4 + cc * 5) : tier(\"default\", p * 4 + c * 18 + cr * 0.4 + cc * 5)"},{"model_name":"seed-2-0-pro-260328","description":"Seed-2.0-pro是一款面向Agent时代的旗舰级通用模型，专为复杂推理与长链任务执行而设计。该模型强化了多模态理解、长上下文推理、结构化生成及工具增强执行能力，在处理复杂指令与多约束执行任务方面表现出色，可稳定应对多步复杂规划、精妙图文推理、视频内容理解及高难度分析等场景。","icon":"ByteDance.Color","tags":"对话,思考,联网搜索","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","BytePlus","default","vip"],"supported_endpoint_types":["openai","openai-response","anthropic"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 128000 ? tier(\"short context\", p * 0.5 + c * 3 + cr * 0.1) : p \u003e 128000 \u0026\u0026 p \u003c= 256000 ? tier(\"long context\", p * 1 + c * 6 + cr * 0.2) : tier(\"default\", p * 0.5 + c * 3 + cr * 0.1)"},{"model_name":"gemini-3.1-flash-tts-preview","description":"Google Gemini 3.1 Flash TTS Preview 是一款文本转语音模型，支持将文本转换为自然流畅的语音，并通过提示词控制语音的情绪、语气和表达风格。适用于有声读物、视频配音、语音播报及多角色对话等场景。","icon":"Gemini.Color","tags":"对话,思考","vendor_id":3,"quota_type":0,"model_ratio":0.5,"model_price":0,"owner_by":"","completion_ratio":4,"audio_ratio":1,"audio_completion_ratio":20,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"gemini-3.5-flash","description":"Gemini 3.5 Flash以标准Flash级别的速度和成本，提供了接近专业级的智能和并行代理执行能力。","icon":"Gemini.Color","tags":"对话,思考,联网搜索","vendor_id":3,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":4,"enable_groups":["Vertex","bingyu_vip","default","Google"],"supported_endpoint_types":["gemini","openai"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 200000 ? tier(\"short context\", p * 1.5 + c * 9 + cr * 0.15 + img * 1.5 + ai * 1.5) : p \u003e 200000 ? tier(\"long context\", p * 1.5 + c * 9 + cr * 0.15 + img * 1.5 + ai * 1.5) : tier(\"default\", p * 1.5 + c * 9 + cr * 0.15 + img * 1.5 + ai * 1.5)"},{"model_name":"gemini-nano-banana-2.1","icon":"Gemini.Color","tags":"文生图,图生图","vendor_id":3,"quota_type":0,"model_ratio":0.75,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"image_ratio":1,"video_ratio":1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai","image-generation"]},{"model_name":"glm-5.2","description":"GLM-5.2 是智谱最新旗舰模型，支持1M上下文及思考长度控制，在coding及agent场景表现优异","icon":"Zhipu.Color","tags":"对话,思考","vendor_id":23,"quota_type":0,"model_ratio":0.5882,"model_price":0,"owner_by":"","completion_ratio":3.500170010201,"cache_ratio":0.25,"enable_groups":["bingyu_vip","default","GLM","GLM_Dep"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"kimi-k3","icon":"Kimi.Avatar","tags":"对话,思考","vendor_id":19,"quota_type":0,"model_ratio":1.5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"enable_groups":["Moonshot_New","bingyu_vip","default","Kimi_Dep","KIMIK3","KIMIK3_Mix","Moonshot_Dep"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"seedance-2-5","description":"Dreamina-Seedance-2.5 新一代多模态视频生成模型，支持文本、图片、视频和音频参考输入，可进行视频生成、视频编辑及视频延长。","icon":"ByteDance.Color","tags":"文生视频,图生视频,视频编辑","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["Seedance_1","bingyu_vip","Seedance"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"seedream-4-5-251128","description":"Seedream 4.5 是基于领先架构的SOTA级多模态图像创作模型，其生成美感、指令遵循、结构完整度、主体保持一致性处于世界头部水平。模型采用同一套架构实现文生图与编辑能力的统一，原生支持文本 、单图和多图输入，并能通过对提示词的深度推理，自动适配最优的图像比例尺寸与生成数量，可一次性连续输出最多 15 张内容关联的图像，支持 4K 超高清输出","icon":"ByteDance.Color","tags":"文生图,图生图","vendor_id":4,"quota_type":1,"model_ratio":0,"model_price":0.04,"owner_by":"","completion_ratio":0,"enable_groups":["default","Seedream","bingyu_vip","BytePlus"],"supported_endpoint_types":["openai","image-generation"]},{"model_name":"seedream-5-0-flash-nsfw","icon":"ByteDance.Color","tags":"文生图,图生图","vendor_id":4,"quota_type":1,"model_ratio":0,"model_price":0.018,"owner_by":"","completion_ratio":0,"enable_groups":["bingyu_vip","BytePlus","default","Seedream"],"supported_endpoint_types":["openai","image-generation"]},{"model_name":"glm-5.3","description":"glm-5.3 是 智谱 AI（Z.ai） 于 2026 年 8 月 14 日 发布的 新一代旗舰级智能体与代码模型。其核心优势集中在智能体编程、长程任务执行和网络安全分析，基于 GLM-5.2 继续进行强化后训练，在减少输出 Token 的同时提升复杂软件工程任务的完成能力。官方测试显示，其在 CyberGym 漏洞发现与验证测试中达到 84.5%，并在多步代码开发、漏洞分析和复杂工程任务上展现出较强能力。模型目前主要面向 AI 编程助手、智能体开发、软件工程、代码审计和网络安全防御等场景。 登录后查看折扣","icon":"Zhipu.Color","tags":"对话,思考","vendor_id":23,"quota_type":0,"model_ratio":0.5882,"model_price":0,"owner_by":"","completion_ratio":3.500170010201,"cache_ratio":0.25,"enable_groups":["default","GLM","GLM_Dep","GLM53_Dep"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"gpt-6-luna","description":"GPT-6 Astra 是 OpenAI 面向复杂端到端任务打造的旗舰级模型，擅长高级推理、软件工程、计算机操作、研究分析、文档创建和多步骤工作流。该模型支持文本与图像输入，具备高性能 Agent 能力，适用于企业自动化和复杂生产任务。","icon":"OpenAI","tags":"对话,思考,联网思考","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":2,"enable_groups":["default","GPT","GPT_K","OpenAI","wy_vip"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 272000 ? tier(\"short context\", p * 0.1 + c * 0.5 + cr * 0.01 + cc * 0.125) : p \u003e 272000 ? tier(\"long context\", p * 0.2 + c * 0.75 + cr * 0.02 + cc * 0.25) : tier(\"default\", p * 0.2 + c * 0.75 + cr * 0.02 + cc * 0.25)"},{"model_name":"claude-opus-4-7","description":"Opus 4.7在高级软件工程领域相较于Opus 4.6实现了显著改进，尤其在处理最棘手任务方面表现突出。用户反馈称，他们能够放心地将最复杂的编码工作——即以往需要严密监督的类型——交由Opus 4.7处理。该版本以严谨性和一致性应对复杂、长期运行的任务，对指令要求精准把控，并通过设计验证机制确保输出结果的准确性后再反馈。\n\n该模型在视觉能力上也显著提升：能以更高分辨率识别图像。在完成专业任务时表现更优雅且富有创意，生成的界面、幻灯片和文档质量更高。尽管其综合能力弱于我们最强大的Claude Mythos Preview模型，但在多项基准测试中均优于Opus 4.6版本","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":2.5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["AwsAnthropic","bingyu_vip","CCAnthropic","default","Anthropic","Anthropic_U"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"claude-sonnet-4-5-20250929","description":"Claude Sonnet 4.5 展示了代理能力的进步，在工具处理、内存管理和上下文处理方面性能得到增强。该模型在代码生成和分析方面表现出显著改进，从识别最佳改进到在重构决策中做出更强判断。它在自主长时程编码任务中表现尤为出色，能够有效规划和执行持续数小时或数天的复杂软件项目，并在整个开发周期中保持一致的性能和可靠性。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":1.5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["Anthropic","Anthropic_U","AwsAnthropic","bingyu_vip","default"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"gpt-image-2","description":"GPT-image-2 是 OpenAI 2026 年 4 月 21 日推出的新一代企业级图像生成模型，现已登陆 Microsoft Foundry，主打 “真实世界智能 + 多语言 + 高分辨率 + 可控批量生成”，面向团队规模化视觉工作流设计。\n它是具备联网搜索、自我校验、多图一致性生成的下一代视觉 AI，知识截止 2025 年 12 月，支持中英日韩印地语等多语言文字渲染，最高4K 分辨率，能让小团队高效批量产出本地化、多尺寸、风格统一的商用图像资产。\n关键能力（精简）\n真实世界智能：知识库更新至 2025 年 12 月，可联网搜索、自检输出、单次提示生成多张一致图像。\n多语言本地化：强化中日韩等语言理解与文字渲染，适配全球市场素材。\n4K 高清输出：支持自定义高分辨率，细节更逼真，满足印刷 / 大屏商用需求。\n企业级可控性：提供智能路由与精细控制，适配批量、多平台、多文化的规模化创意工作流。","icon":"OpenAI","tags":"文生图,图生图","vendor_id":6,"quota_type":0,"model_ratio":2.5,"model_price":0,"owner_by":"","completion_ratio":2,"cache_ratio":0.25,"image_ratio":1.6,"enable_groups":["xy_vip","default","GPT_Image","unal_vip","wy_vip","bingyu_vip","GPT","oai_image","OpenAI"],"supported_endpoint_types":["openai","images-edits","image-generation"]},{"model_name":"wan2.7-t2v","description":"Wan2.7-T2V，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。","icon":"Qwen.Color","tags":"文生视频","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","Wan","bingyu_vip"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"gpt-image-1","description":"GPT-image-1 是 OpenAI 于 2025 年 4 月推出的图像生成模型。支持文本到图像、图像到图像生成及智能编辑，能精准理解复杂指令，生成细节丰富、风格多样的高质量图像。","icon":"OpenAI","tags":"文生图,图生图","vendor_id":6,"quota_type":0,"model_ratio":2.5,"model_price":0,"owner_by":"","completion_ratio":2,"image_ratio":2,"enable_groups":["unal_vip","xy_vip","default","GPT","GPT_Image","OpenAI"],"supported_endpoint_types":["image-generation","openai"]},{"model_name":"gpt-realtime-2.1-mini","icon":"OpenAI","tags":"实时语音","vendor_id":6,"quota_type":0,"model_ratio":0.3,"model_price":0,"owner_by":"","completion_ratio":4,"image_ratio":1.333333,"audio_ratio":16.666666,"audio_completion_ratio":2,"enable_groups":["GPT_K","OpenAI","wy_vip","default","GPT"],"supported_endpoint_types":["openai","openai-realtime"]},{"model_name":"gemini-2.5-pro","description":"Gemini-2.5-Pro 是一款强大的多模态 AI 模型，专为处理复杂任务而设计。它支持文本、图像和音频等输入，具备出色的推理和生成能力，适用于需要高精度和多步骤逻辑的应用场景，如数据分析、内容创作和高级编程辅助。\n输入： 文本、 代码、 图片、 音频、 视频\n输出： 文本","icon":"Gemini.Color","tags":"对话,思考","vendor_id":3,"quota_type":0,"model_ratio":0.625,"model_price":0,"owner_by":"","completion_ratio":8,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"],"billing_mode":"tiered_expr","billing_expr":"len \u003c= 200000 ? tier(\"short context\", p * 1.25 + c * 10 + cr * 0.125) : len \u003e 200000 ? tier(\"long context\", p * 2.5 + c * 15 + cr * 0.25) : tier(\"第3档\", p * 2.5 + c * 15 + cr * 0.25)"},{"model_name":"gemini-3.1-flash-image-preview","description":"Gemini 3.1 Flash Image的图像生成能力达到了主流水平。它提升了图像质量，增强了视觉理解能力，能够更好地渲染多种语言的长文本段落，并提高了事实准确性。该模型还能生成分辨率在512（0.25MP）至4K（16MP）之间的图像。","icon":"Gemini.Color","tags":"文生图,图生图","vendor_id":3,"quota_type":0,"model_ratio":0.25,"model_price":0,"owner_by":"","completion_ratio":6,"image_ratio":1,"enable_groups":["Google","Vertex","bingyu_vip","default"],"supported_endpoint_types":["gemini","openai","image-generation"]},{"model_name":"gemini-3.1-flash-lite","icon":"Gemini.Color","tags":"对话,思考,联网搜索","vendor_id":3,"quota_type":0,"model_ratio":0.125,"model_price":0,"owner_by":"","completion_ratio":6,"cache_ratio":0.1,"image_ratio":1,"audio_ratio":2,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"seedream-5-0-lite-nsfw","description":"BytePlus Seedream 5.0 Lite 图片生成模型，适用于高质量文生图、图生图和多图融合场景。该接入点用于内部 NSFW 内容场景，请遵守平台和法律规定。","icon":"ByteDance.Color","tags":"文生图,图生图","vendor_id":4,"quota_type":1,"model_ratio":0,"model_price":0.035,"owner_by":"","completion_ratio":0,"enable_groups":["Seedream","bingyu_vip","BytePlus","default"],"supported_endpoint_types":["openai","image-generation"]},{"model_name":"veo-3.1-generate-preview","quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default"],"supported_endpoint_types":["openai"]},{"model_name":"hy4-preview","icon":"Hunyuan.Color","tags":"对话,思考","vendor_id":39,"quota_type":0,"model_ratio":0.42855,"model_price":0,"owner_by":"","completion_ratio":3.0001166725,"cache_ratio":0.050052502625,"enable_groups":["bingyu_vip","default","Hunyuan"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"wan2.7-i2v","description":"万相2.7-图生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。","icon":"Qwen.Color","tags":"图生视频","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["Wan","bingyu_vip","default"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"happyhorse-1.1-i2v","description":"HappyHorse-1.1-I2V支持图生视频，进一步提升画面质感、动态表现与跨片段一致性。模型能够更精准地理解输入图像并延续创作意图，在人物皮肤质感、ID跨片段保持、动作流畅度、文字渲染稳定性以及音画同步上带来显著改善，输出更真实自然、细节丰富且一致性更高的高质量视频。","icon":"HappyHorse","tags":"图生视频","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","HappyHouse"],"supported_endpoint_types":["openai","video"]},{"model_name":"happyhorse-1.1-r2v","description":"HappyHorse-1.1-R2V支持参考生视频，进一步提升主体、场景风格与画面一致性的稳定保持。模型最多支持9张参考图片输入，能够更精准理解并延续创作意图，在人物、场景、风格和镜头表现上带来更强的可控性与表现力。","icon":"HappyHorse","tags":"文生视频,图生视频","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","HappyHouse"],"supported_endpoint_types":["openai","video"]},{"model_name":"claude-opus-5-5","description":"Opus 5代表Opus系列实现跨越式升级：代码编写能力更强，智能代理性能更出色，专业任务处理水平更高。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":2,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.05,"create_cache_ratio":1.25,"enable_groups":["Anthropic","Anthropic_U","AwsAnthropic","bingyu_vip","CCAnthropic","default"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"gemini-3.8-flash","description":"Gemini 3.8 Flash 是一款面向通用场景的高性能推理模型，专注于软件工程、智能体工作流、多步骤推理和复杂知识任务。模型支持文本、图片、音频和视频输入，具备百万 token 长上下文和可调节的思考强度，在速度、成本与智能之间实现平衡。","icon":"Gemini.Color","tags":"对话,思考","vendor_id":3,"quota_type":0,"model_ratio":0.375,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"seedream-5-0-pro-NSFW","description":"BytePlus Seedream 5.0 Pro 图片生成模型，支持文生图和图生图，适用于高质量图片创作、图片编辑和多参考图生成。","icon":"ByteDance.Color","tags":"文生图,图生图","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","BytePlus","default","Seedream","Seedream5Pro"],"supported_endpoint_types":["openai","image-generation"]},{"model_name":"seedance-1-5-pro-251215-nsfw","description":"Seedance 1.5 pro 作为全球领先的视频生成模型，可生成音画高精同步的视频内容。支持多人多语言对白，全面覆盖环境音、动作音、合成音、乐器音、背景音及人声，支持首尾帧，实现影视级叙事效果，满足影视、漫剧、电商及广告领域的高阶创作需求。","icon":"ByteDance.Color","tags":"文生视频","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"video_ratio":0,"enable_groups":["default","vip","BytePlus"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"veo-3.1-fast-generate-preview","quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default"],"supported_endpoint_types":["openai"]},{"model_name":"gpt-image-2.5-sunburst","description":"GPT Image 2.5 Sunburst 是 OpenAI 的高质量图像生成与编辑模型，支持根据文本提示生成图片，并可结合参考图片进行编辑、变体创作和视觉内容生成。","icon":"OpenAI","tags":"文生图,图生图","vendor_id":6,"quota_type":0,"model_ratio":2.5,"model_price":0,"owner_by":"","completion_ratio":2,"cache_ratio":0.25,"image_ratio":1.6,"enable_groups":["wy_vip","bingyu_vip","default","GPT","GPT_Image","oai_image","OpenAI"],"supported_endpoint_types":["openai","openai-images","openai-images-edits"]},{"model_name":"deepseek-v4.1-flash","icon":"DeepSeek.Color","tags":"对话,思考","vendor_id":5,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["DeepSeek","default"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"],"billing_mode":"tiered_expr","billing_expr":"(tier(\"base\", p * 0.15 + c * 0.6 + cr * 0.003)) * (hour(\"Asia/Shanghai\") \u003e= 9 \u0026\u0026 hour(\"Asia/Shanghai\") \u003c 12 \u0026\u0026 weekday(\"Asia/Shanghai\") \u003c= 5 ? 2 : 1) * (hour(\"Asia/Shanghai\") \u003e= 14 \u0026\u0026 hour(\"Asia/Shanghai\") \u003c 18 \u0026\u0026 weekday(\"Asia/Shanghai\") \u003c= 5 ? 2 : 1)"},{"model_name":"gpt-5.1-codex","description":"GPT-5.1-codex：在 GPT-5-codex 的基础上进一步增强了开发者工具链，强化推理框架，支持上下文感知的代码分析与生成，工具处理能力更强，多模态交互让开发体验更丰富。结合 Foundry 的企业级安全治理，助力自动化代码生成、审查、重构和 Bug 检测，加速开发周期。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.625,"model_price":0,"owner_by":"","completion_ratio":8,"cache_ratio":0.104,"enable_groups":["default","OpenAI","unal_vip","xy_vip"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"happyhorse-1.1-t2v","description":"HappyHorse-1.1-T2V支持文生视频，进一步提升文本语义理解、镜头调度与动态生成表现。模型能够更精准地还原创作意图，在人物动作、场景氛围、视觉美感和物理运动上生成更流畅自然、细节丰富且一致性更高的高质量视频。","icon":"HappyHorse","tags":"文生视频","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","HappyHouse"],"supported_endpoint_types":["openai","video"]},{"model_name":"qwen3.7-flash","description":"阿里通义 Qwen3.7 系列的轻量级（Flash）版本，定位在 Plus/Max 之下，于 2026 年 7 月发布。原生视觉语言模型，支持文本、图像、视频输入，上下文窗口约 99 万 token，单次输出上限 6.4 万 token。相较 Qwen3.6-Flash，在多模态理解与智能体（agent）执行能力上全面升级，尤其擅长物体识别、空间理解等真实视觉感知任务，适合多模态智能体、视觉编程、搜索和电脑操作场景。定价极低（输入约 $0.03/百万 token，输出约 $0.13/百万 token），是目前最便宜的视觉能力模型之一。","icon":"Qwen.Color","tags":"对话,思考,联网搜索","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","default","Qwen"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 32000 ? tier(\"limit context\", p * 0.0294 + c * 0.1176 + cr * 0.0059) : p \u003e 32000 \u0026\u0026 p \u003c= 256000 ? tier(\"short context\", p * 0.0882 + c * 0.3529 + cr * 0.0176) : p \u003e 256000 \u0026\u0026 p \u003c= 1000000 ? tier(\"long context\", p * 0.1765 + c * 0.7059 + cr * 0.0353) : tier(\"default\", p * 0.1765 + c * 0.7059 + cr * 0.0353)"},{"model_name":"veo-3.1-generate-001","description":"Veo 3.1 for Generation 将创意转化为引人注目的创意 使用 Google 高级视频生成模型的视频叙事。Veo 有能力 根据文本提示生成带有音频的视频，或使用 文本指导。","icon":"Gemini.Color","tags":"文生视频,图生视频","vendor_id":3,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["Google","Vertex","bingyu_vip","default"],"supported_endpoint_types":["gemini","openai","video-generation"]},{"model_name":"grok-imagine-image","description":"grok-imagine-image——xAI 推出的、基于自研 Aurora 自回归引擎的文生图 / 图编辑模型，是 Grok Imagine 多模态套件的核心图像分支，面向创作者与开发者提供高精度视觉生成能力。","icon":"Grok","tags":"文生图,图生图","vendor_id":14,"quota_type":1,"model_ratio":0,"model_price":0.02,"owner_by":"","completion_ratio":0,"enable_groups":["default","Grok"],"supported_endpoint_types":["openai","openai-response","openai-images"]},{"model_name":"gpt-5.6-luna","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":6,"enable_groups":["AzureGPT","default","GPT","GPT_K","OpenAI","wy_vip"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 272000 ? tier(\"short context\", p * 0.2 + c * 1.2 + cr * 0.02 + cc * 0.25) : p \u003e 272000 ? tier(\"long context\", p * 0.4 + c * 1.8 + cr * 0.04 + cc * 0.5) : tier(\"default\", p * 0.4 + c * 1.8 + cr * 0.04 + cc * 0.5)"},{"model_name":"gemini-2.5-flash-image","description":"Gemini-2.5-Flash-Image 是一款专注于图像预览的轻量级 AI 模型。它能够快速处理和展示图像内容，支持多模态输入，尤其适合需要快速响应和低延迟的应用场景，如实时图像处理和内容预览。\n输入：文本，图片,音频\n输出：文本，图片","icon":"Gemini.Color","tags":"文生图,图生图","vendor_id":3,"quota_type":0,"model_ratio":0.15,"model_price":0,"owner_by":"","completion_ratio":8.333333,"enable_groups":["Vertex","bingyu_vip","default","Google"],"supported_endpoint_types":["gemini","openai","image-generation"]},{"model_name":"mimo-v2.6-flash","icon":"XiaomiMiMo","tags":"对话,思考","vendor_id":36,"quota_type":0,"model_ratio":0.07355,"model_price":0,"owner_by":"","completion_ratio":1.999320190347,"cache_ratio":0.019714479946,"enable_groups":["default","MIMO"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"seedance-2-0-fast-NSFW","description":"Seedance-2.0-fast是新一代的多模态视频创作模型，它继承了Seedance-2.0的核心功能和优势，同时速度更快。","icon":"ByteDance.Color","tags":"文生视频,图生视频","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","Seedance","Seedance_1"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"seedance-2-0-NSFW","description":"Seedance 2.0 NSFW","icon":"ByteDance.Color","tags":"文生视频,图生视频","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","Seedance","Seedance_1"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"claude-fable-5","description":"Claude Fable 5 是 Anthropic 全新 Claude 5 系列中的首个模型,属于新设立的 \"Mythos 级\"(Mythos-class)模型层级,该层级在能力上位于 Claude Opus 之上。\nClaude Fable 5 与 Claude Mythos 5 基于相同的底层模型,区别在于 Claude Fable 5 加入了额外的安全措施,因此可以面向公众开放使用。它的 API 模型字符串为 claude-fable-5,可在与其他 Claude 模型相同的各类入口中使用(网页/桌面/移动端、API 等)。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["Anthropic","Anthropic_U","bf_vip","CCAnthropic","default"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"claude-opus-5","description":"Opus 5代表Opus系列实现跨越式升级：代码编写能力更强，智能代理性能更出色，专业任务处理水平更高。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":2.5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["CCAnthropic","default","Anthropic","Anthropic_U","AwsAnthropic","bingyu_vip"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"gpt-4.1-mini","description":"GPT-4.1-mini 是 GPT-4.1 系列中的轻量级版本，专为低成本和低延迟的应用场景设计。它继承了 GPT-4o-mini 的功能，支持文本和视觉输入，能够处理多种任务，如串联或并行调用多个模型（例如调用多个API）、传递大量上下文（例如完整代码库或对话历史）以及通过快速实时文本响应与客户互动（例如客服聊天机器人）。此外，它还支持高达128k的上下文窗口，适用于需要快速响应和成本效益的应用，如实时客服和多模型调用场景。\n输入：文本和图像\n输出：文本","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.2,"model_price":0,"owner_by":"","completion_ratio":4,"cache_ratio":0.25,"enable_groups":["OpenAI","wy_vip","default","GPT","GPT_K"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"text-embedding-3-large","description":"text-embedding-3-large 是 OpenAI 的高性能文本嵌入模型，最大输入 8191 令牌，最高可生成 3072 维向量，还支持灵活降维。它用套娃表示学习技术训练，MTEB 等基准测试表现优异，多语言理解能力强，能高效完成语义搜索、文档聚类等任务，适配长文档处理，适合高精准度需求的企业级场景。","icon":"OpenAI","tags":"嵌入","vendor_id":6,"quota_type":0,"model_ratio":0.0715,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["OpenAI","wy_vip","default","GPT","GPT_K"],"supported_endpoint_types":["openai","embeddings"]},{"model_name":"claude-opus-4-8","description":"基于 Opus 4.7 升级，在编码、Agent 任务与专业协作场景更稳定可靠","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":2.5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["Anthropic","Anthropic_U","AwsAnthropic","bingyu_vip","CCAnthropic","default"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"gpt-4o-mini-tts","description":"OpenAI 基于 GPT-4o mini 推出的新一代可指令文本转语音（TTS）模型，支持通过自然语言控制语气、情绪、语速、语调与口音等表达风格。内置 13 种预置音色，适合客服播报、有声内容、语音助手等高性价比生产场景。","icon":"OpenAI","tags":"对话,思考","vendor_id":6,"quota_type":0,"model_ratio":0.3,"model_price":0,"owner_by":"","completion_ratio":20,"enable_groups":["default","GPT","GPT_K","OpenAI","wy_vip"],"supported_endpoint_types":["openai","openai-speech"]},{"model_name":"happyhorse-1.0-t2v","description":"HappyHorse-1.0-T2V支持文生视频，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。","icon":"Qwen.Color","tags":"文生视频","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["HappyHouse","default"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"veo-3.0-fast-generate-001","quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default"],"supported_endpoint_types":["openai"]},{"model_name":"gpt-4.1","description":"GPT-4.1 是 GPT-4o 模型家族的最新迭代版本，专为提升编程和指令遵循能力而设计，能够更有效地处理复杂的编程和技术问题。该模型将上下文令牌限制提高到最多100万输入令牌，并为小上下文（128k）和大上下文（最多100万）输入提供单独计费。它继承了 GPT-4o 的多种功能，包括文本和图像处理、JSON 模式、并行函数调用、增强的准确性和响应速度，以及对复杂结构化输出的支持。此外，它在非英语语言和视觉任务方面表现出色，与 GPT-4 Turbo with Vision 在英语文本和编程任务上具有同等性能。\n输入：文本和图像\n输出：文本","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":1,"model_price":0,"owner_by":"","completion_ratio":4,"cache_ratio":0.25,"enable_groups":["OpenAI","wy_vip","default","GPT","GPT_K"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"gemini-omni-1.1-flash-preview","description":"Gemini Omni 1.1 Flash Preview 是一款高性能多模态视频生成与编辑模型，支持通过文本或图片生成视频，并可进行自然语言驱动的对话式视频编辑。","icon":"Gemini.Color","tags":"文生视频,图生视频,视频编辑","vendor_id":3,"quota_type":0,"model_ratio":0.75,"model_price":0,"owner_by":"","completion_ratio":6,"image_ratio":1,"audio_ratio":1,"video_ratio":1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai","openai-video","google-interactions"]},{"model_name":"seedream-5-0-flash-260915","description":"Dola Seedream 5.0 Flash 是 BytePlus 推出的高性能图像生成与编辑模型，面向低时延、成本敏感的创作场景。支持文生图、单图/多参考图生图、基于坐标与标记的精准局部编辑、图层拆分及多语言文字生成，可将单张图片拆解为底图和最多 16 个可独立编辑图层。适用于电商素材、营销海报、视觉设计、图片编辑与批量内容创作等场景。","icon":"ByteDance.Color","tags":"文生图,图生图","vendor_id":4,"quota_type":1,"model_ratio":0,"model_price":0.018,"owner_by":"","completion_ratio":0,"enable_groups":["default","Seedream","bingyu_vip","BytePlus"],"supported_endpoint_types":["openai","image-generation"]},{"model_name":"veo-3.1-fast-generate-001","description":"Veo 3.1 Generation 利用 Google 先进的视频生成模型，将创意转化为引人入胜的视频叙事。Veo 能够根据文本提示生成带有音频的视频，或根据文本引导生成动画图像。","icon":"Gemini.Color","tags":"文生视频,图生视频","vendor_id":3,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai","video-generation"]},{"model_name":"wan2.7-r2v","description":"Wan2.7-R2V，更加稳定的角色、道具与场景参考，支持最大5个图/视频混合参考，支持音频音色参考，搭配基础能力升级实现更强表演能力。","icon":"Qwen.Color","tags":"参考生视频","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["Wan","bingyu_vip","default"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"grok-4.6","description":"Grok 4.6 是 xAI 面向编程、智能代理任务和知识工作的旗舰模型，支持文本和图像输入、文本输出、函数调用、结构化输出及可配置推理。","icon":"Grok","tags":"对话,思考","vendor_id":14,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","Grok"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"len \u003c 200000 ? tier(\"short context\", p * 2 + c * 6 + cr * 0.5) : len \u003e 200000 ? tier(\"long context\", p * 4 + c * 12 + cr * 1) : tier(\"default档\", p * 4 + c * 12 + cr * 1)"},{"model_name":"mimo-v2.6-pro","icon":"XiaomiMiMo","tags":"对话,思考","vendor_id":36,"quota_type":0,"model_ratio":0.2206,"model_price":0,"owner_by":"","completion_ratio":2,"cache_ratio":0.008386219402,"enable_groups":["default","MIMO"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"gpt-6-astra","description":"GPT-6 Astra 是 OpenAI 面向复杂端到端任务打造的旗舰级模型，擅长高级推理、软件工程、计算机操作、研究分析、文档创建和多步骤工作流。该模型支持文本与图像输入，具备高性能 Agent 能力，适用于企业自动化和复杂生产任务。","icon":"OpenAI","tags":"对话,思考,联网思考","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":2,"enable_groups":["bingyu_vip","default","GPT","GPT_K","OpenAI","wy_vip","AzureGPT"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"len \u003c= 272000 ? tier(\"short context\", p * 10 + c * 50 + cr * 1 + cc * 12.5) : len \u003e 272000 ? tier(\"long context\", p * 20 + c * 75 + cr * 2 + cc * 25) : tier(\"default\", p * 20 + c * 75 + cr * 2 + cc * 25)"},{"model_name":"deepseek-v4-flash-0731","icon":"DeepSeek.Color","tags":"对话,思考","vendor_id":5,"quota_type":0,"model_ratio":0.2272,"model_price":0,"owner_by":"","completion_ratio":2.996919014085,"cache_ratio":0.100352112676,"enable_groups":["bingyu_vip","DeepSeek","default"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"deepseek-v4.1-flash-1017","icon":"DeepSeek.Color","tags":"对话,思考","vendor_id":5,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","bingyu_vip","DeepSeek"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"],"billing_mode":"tiered_expr","billing_expr":"(tier(\"base\", p * 0.15 + c * 0.6 + cr * 0.003)) * (hour(\"Asia/Shanghai\") \u003e= 9 \u0026\u0026 hour(\"Asia/Shanghai\") \u003c 12 \u0026\u0026 weekday(\"Asia/Shanghai\") \u003c= 5 ? 2 : 1) * (hour(\"Asia/Shanghai\") \u003e= 14 \u0026\u0026 hour(\"Asia/Shanghai\") \u003c 18 \u0026\u0026 weekday(\"Asia/Shanghai\") \u003c= 5 ? 2 : 1)"},{"model_name":"seed-2-0-code-preview-260328","icon":"ByteDance.Color","tags":"对话,思考,联网搜索","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","BytePlus","default","vip"],"supported_endpoint_types":["openai","openai-response","anthropic"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 128000 ? tier(\"short context\", p * 0.5 + c * 3 + cr * 0.1) : p \u003e 128000 \u0026\u0026 p \u003c= 256000 ? tier(\"long context\", p * 1 + c * 6 + cr * 0.2) : tier(\"default\", p * 1 + c * 6 + cr * 0.2)"},{"model_name":"veo-3.0-generate-001","quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default"],"supported_endpoint_types":["openai"]},{"model_name":"gpt-5.4-nano","description":"GPT‑5.4  是 OpenAI 最新推出、面向专业工作场景的高性能模型，集成了更强的推理、编程与智能体能力，能更精准高效地完成复杂任务，减少反复沟通。它支持可查看、可中途调整的思考过程，搭配更强的网页搜索与长上下文理解，同时原生具备计算机操作能力，上下文长度可达 1M Token，推理更省 Token、响应更快，适合高强度、高复杂度的专业工作。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.1,"model_price":0,"owner_by":"","completion_ratio":6.25,"cache_ratio":0.1,"enable_groups":["bingyu_vip","default","OpenAI","unal_vip","xy_vip","AzureGPT"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"deepseek-v4-pro-0813","icon":"DeepSeek.Color","tags":"对话,思考","vendor_id":5,"quota_type":0,"model_ratio":0.6809,"model_price":0,"owner_by":"","completion_ratio":3.001028051109,"cache_ratio":0.100455279777,"enable_groups":["DeepSeek","default","bingyu_vip"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"seed-2-1-turbo-260628","icon":"ByteDance.Color","tags":"对话,思考","vendor_id":4,"quota_type":0,"model_ratio":0.25,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.2,"enable_groups":["bingyu_vip","BytePlus","default","vip"],"supported_endpoint_types":["openai","openai-response","anthropic"]},{"model_name":"seedance-2-5-NSFW","description":"Seedance 2.5 NSFW 新一代多模态视频生成模型，支持文本、图片、视频和音频参考输入，可进行视频生成、视频编辑及视频延长。","icon":"ByteDance.Color","tags":"文生视频,图生视频,视频编辑","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","Seedance","Seedance_1"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"gpt-realtime-1.5","icon":"OpenAI","tags":"实时语音","vendor_id":6,"quota_type":0,"model_ratio":2,"model_price":0,"owner_by":"","completion_ratio":4,"image_ratio":1.25,"audio_ratio":8,"audio_completion_ratio":2,"enable_groups":["GPT_K","OpenAI","wy_vip","default","GPT"],"supported_endpoint_types":["openai","openai-realtime"]},{"model_name":"grok-imagine-image-2.0","description":"grok-imagine-image-2.0 是 xAI 于 2026 年 8 月推出的新一代图像生成与编辑模型，主要面向高质量视觉创作。其优势集中在更强的提示词理解、图像编辑控制、真实感表现和创意生成能力，支持文本生成图像以及基于图像的编辑，并可输出最高 2K 分辨率。相比此前版本，更适合进行人物与场景创作、广告素材、电商视觉、社交媒体内容、产品视觉及创意图像编辑。xAI 官方目前已将 Grok Imagine 图像能力作为统一视觉创作接口的一部分提供。 登录后查看折扣","icon":"Grok","tags":"文生图,图生图","vendor_id":14,"quota_type":1,"model_ratio":0,"model_price":0.08,"owner_by":"","completion_ratio":0,"enable_groups":["default","Grok"],"supported_endpoint_types":["openai","openai-response","openai-images"]},{"model_name":"claude-sonnet-5-5","description":"Claude Sonnet 第五代主力模型，在 Sonnet 4.6 基础上全面升级编码、工具调用与长程 Agent 能力，兼顾性能与成本，是日常开发与 Agent 工作流的首选平衡型旗舰。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":1,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.05,"create_cache_ratio":1.25,"enable_groups":["AwsAnthropic","bingyu_vip","CCAnthropic","default","Anthropic","Anthropic_U"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"gemini-3-flash-preview","description":"Gemini 3 Flash Preview 旨在以显著的速度和价值提供强大的代理能力（接近专业水平），非常适合进行多轮对话，并能与您的编码代理来回协作而不打断流程。与 2.5 Flash 相比，它在各方面都有显著改进。是强大的代理和编码模型。它具有 1M token 的上下文窗口，并拥有最佳的多模态理解能力。","icon":"Gemini.Color","tags":"对话,思考,联网搜索","vendor_id":3,"quota_type":0,"model_ratio":0.25,"model_price":0,"owner_by":"","completion_ratio":6,"cache_ratio":0.1,"audio_ratio":2,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"gpt-5-mini","description":"GPT-5-mini 是 GPT-5 系列中的轻量级版本，专为成本敏感型应用设计，能够提供低成本且快速的体验，例如实时客服代理。它支持最小化推理、新的详细程度设置以及“customs”工具，用于原始文本输出。此外，它还支持新的“allowed tools”工具选择，允许在工具选择中指定多个工具，而不仅仅是单一工具。它还具备“preamble”支持，使模型在调用工具之前能够进行“思考”，这一功能始终启用并通过提示进行控制。GPT-5-mini 支持多模态输入、实时流式处理以及全面的工具支持，为用户带来更智能、更动态的体验。\n输入：文本和图像\n输出：文本","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.125,"model_price":0,"owner_by":"","completion_ratio":8,"cache_ratio":0.1,"enable_groups":["default","OpenAI","unal_vip","xy_vip"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"gpt-realtime-2.1","icon":"OpenAI","tags":"实时语音","vendor_id":6,"quota_type":0,"model_ratio":2,"model_price":0,"owner_by":"","completion_ratio":6,"image_ratio":1.25,"audio_ratio":8,"audio_completion_ratio":2,"enable_groups":["default","GPT","GPT_K","OpenAI","wy_vip"],"supported_endpoint_types":["openai","openai-realtime"]},{"model_name":"glm-5.3-flash","icon":"Zhipu.Color","tags":"对话,思考","vendor_id":23,"quota_type":0,"model_ratio":0.0588,"model_price":0,"owner_by":"","completion_ratio":3.501700680272,"cache_ratio":0.287414965986,"enable_groups":["default","GLM_Dep"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"claude-fable-5-1","description":"高阶推理与长程智能体模型，Anthropic Claude Fable 5.1 系列模型，核心优势集中在复杂编码、长程 Agent 任务、多步骤研究与企业知识工作。支持文本与图像输入、文本输出，具备 1M tokens 上下文窗口与 128K tokens 最大输出，并默认启用自适应思考能力，适合处理需要持续规划、工具协作和长链路执行的高难度任务。适合复杂代码开发、大型迁移、自动化 Agent、深度研究、文档 / 表格 / 幻灯片处理、企业分析及高难度知识工作。Anthropic 官方将 Claude Fable 5.1 定位为面向 demanding reasoning 与 long-horizon agentic work 的最新可用模型。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.025,"create_cache_ratio":1.25,"enable_groups":["bf_vip","CCAnthropic","default","Anthropic","Anthropic_U"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"gemini-2.5-flash-lite","description":"Gemini-2.5-Flash-Lite 是个很厉害的“小帮手”，它特别快，能瞬间回答问题，就像闪电一样。它还能处理文字、图片，甚至音频，功能特别多。和其他模型比，它更轻巧，用起来成本低，特别适合需要快速、低成本解决问题的时候。\n输入：文本、图片、视频，音频\n输出：文本","icon":"Gemini.Color","tags":"对话,思考","vendor_id":3,"quota_type":0,"model_ratio":0.05,"model_price":0,"owner_by":"","completion_ratio":4,"cache_ratio":0.1,"audio_ratio":3,"enable_groups":["default","Google","Vertex","bingyu_vip"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"qwen3.7-plus","description":"Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。","icon":"Qwen.Color","tags":"对话,思考,联网搜索","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["Qwen","bingyu_vip","default"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 256000 ? tier(\"short context\", p * 0.2941 + c * 1.1765 + cr * 0.0588) : p \u003e 256000 \u0026\u0026 p \u003c= 1000000 ? tier(\"long context\", p * 0.8824 + c * 3.5294 + cr * 0.1765) : tier(\"default\", p * 0.7059 + c * 2.8235 + cr * 0.1412)"},{"model_name":"seedance-2-0","description":"Seedance 2.0 作为全球领先的视频生成模型，可生成音画高精同步的视频内容。支持多人多语言对白，全面覆盖环境音、动作音、合成音、乐器音、背景音及人声，支持首尾帧，实现影视级叙事效果，满足影视、漫剧、电商及广告领域的高阶创作需求。","icon":"ByteDance.Color","tags":"文生视频,图生视频","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","Seedance","Seedance_1"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"gpt-5.4-pro","description":"GPT‑5.4-Pro 是 OpenAI 最新推出、面向专业工作场景的高性能模型，集成了更强的推理、编程与智能体能力，能更精准高效地完成复杂任务，减少反复沟通。它支持可查看、可中途调整的思考过程，搭配更强的网页搜索与长上下文理解，同时原生具备计算机操作能力，上下文长度可达 1M Token，推理更省 Token、响应更快，适合高强度、高复杂度的专业工作。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":6,"enable_groups":["AzureGPT","bingyu_vip","default","GPT","wy_vip","GPT_K","OpenAI","unal_vip","xy_vip"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 272000 ? tier(\"short context\", p * 30 + c * 180) : p \u003e 272000 ? tier(\"long context\", p * 60 + c * 270) : tier(\"default\", p * 60 + c * 270)"},{"model_name":"gemini-3.7-flash","description":"Gemini 3.7 Flash 是 Google Gemini 3 系列的高性能、低延迟多模态推理模型，适用于代码生成与重构、复杂问题分析、多步骤 Agent 工作流、工具调用以及文本、图片、音频和视频理解。该模型支持百万级上下文窗口，并可通过不同思考等级在响应速度与推理深度之间进行平衡。","icon":"Gemini.Color","tags":"对话,思考","vendor_id":3,"quota_type":0,"model_ratio":0.375,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"lyria-3-pro-preview","description":"lyria-3-pro-preview 是 Google 推出的旗舰级音乐生成模型，可根据文本提示或图片生成完整音乐作品。模型支持复杂的歌曲结构，包括前奏、主歌、副歌和桥段，并能够生成带人声和歌词的音乐。它适用于影视配乐、广告音乐、游戏音乐、短视频背景音乐和音乐创作等场景，可生成高质量的 48 kHz 立体声音频。","icon":"Gemini.Color","tags":"多模态,语音","vendor_id":3,"quota_type":1,"model_ratio":0,"model_price":0.08,"owner_by":"","completion_ratio":0,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai","google-interactions"]},{"model_name":"deepseek-v4-pro","description":"deepseek-v4-pro 旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。","icon":"DeepSeek.Color","tags":"对话,思考","vendor_id":5,"quota_type":0,"model_ratio":0.9081,"model_price":0,"owner_by":"","completion_ratio":2.250192710054,"cache_ratio":0.083415923356,"enable_groups":["DeepSeek","default"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"claude-opus-4-6","description":"Claude Opus 4.6 是 Anthropic 最智能的模型，为生产代码、复杂代理和复杂办公任务设定了新标准。Opus 4.6 提供最先进的编码能力，性能比之前的 Opus 级智能降低了三分之一。它在需要从多个数据源获取信息、处理模糊性并无需人工指导进行权衡推理的复杂、长时间运行的任务中表现优异。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":2.5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["Anthropic_U","AwsAnthropic","bingyu_vip","CCAnthropic","default","Anthropic"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"seedream-5-0-260128","description":"Doubao-Seedream-5.0 是字节跳动发布的最新图像创作模型。该模型首次搭载联网检索功能，能融合实时网络信息，提升生图时效性。同时，模型的聪明度进一步升级，能够精准解析复杂指令和视觉内容。此外，模型在世界知识广度、参考一致性及专业场景生成质量上均有增强，可更好地满足企业级视觉创作需求。","icon":"ByteDance.Color","tags":"文生图,图生图","vendor_id":4,"quota_type":1,"model_ratio":0,"model_price":0.035,"owner_by":"","completion_ratio":0,"enable_groups":["bingyu_vip","BytePlus","default","Seedream"],"supported_endpoint_types":["openai","image-generation"]},{"model_name":"gemini-4-argon","description":"Gemini 3.8 Flash 是一款面向通用场景的高性能推理模型，专注于软件工程、智能体工作流、多步骤推理和复杂知识任务。模型支持文本、图片、音频和视频输入，具备百万 token 长上下文和可调节的思考强度，在速度、成本与智能之间实现平衡。","icon":"Gemini.Color","tags":"对话,思考","vendor_id":3,"quota_type":0,"model_ratio":1,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.05,"image_ratio":1,"audio_ratio":1,"video_ratio":1,"enable_groups":["Vertex","bingyu_vip","default","Google"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"gemini-3.1-flash-lite-image","icon":"Gemini.Color","tags":"文生图,图生图","vendor_id":3,"quota_type":0,"model_ratio":0.125,"model_price":0,"owner_by":"","completion_ratio":6,"cache_ratio":0.1,"image_ratio":1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai","image-generation"]},{"model_name":"gemini-3.8-flash-tts","description":"Google Gemini 3.8 Flash TTS 是一款高保真文本转语音模型，支持将文本转换为自然、富有表现力的语音，并通过提示词控制语音风格、情绪和语气。支持多语言及多说话人语音生成，适用于有声读物、视频配音、语音助手、播客和对话式语音应用。","icon":"Gemini.Color","tags":"对话,思考","vendor_id":3,"quota_type":0,"model_ratio":0.25,"model_price":0,"owner_by":"","completion_ratio":4,"audio_ratio":1,"audio_completion_ratio":18,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"gpt-image-2.5-flare","description":"图像生成与编辑模型，OpenAI GPT Image 2.5 系列高速版本，针对日常图像创作进行了速度与质量优化。相比 GPT Image 2，生成延迟最高降低约 50%，同时提升参考图保真度、自然光影、纹理细节及多轮编辑一致性，适合社媒内容、产品视觉、广告素材、视觉搜索及高频批量生成。","icon":"OpenAI","tags":"文生图,图生图","vendor_id":6,"quota_type":0,"model_ratio":2.5,"model_price":0,"owner_by":"","completion_ratio":2,"cache_ratio":0.25,"image_ratio":1.6,"enable_groups":["wy_vip","bingyu_vip","default","GPT","GPT_Image","oai_image","OpenAI"],"supported_endpoint_types":["openai","openai-images","openai-images-edits"]},{"model_name":"veo-3.1-lite-generate-001","description":"Veo 3.1 for Generation 将创意转化为引人注目的创意 使用 Google 高级视频生成模型的视频叙事。Veo 有能力 根据文本提示生成带有音频的视频，或使用 文本指导。","icon":"Gemini.Color","tags":"文生视频,图生视频","vendor_id":3,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai","video-generation"]},{"model_name":"gpt-6.1-sol","description":"GPT‑6.1 Sol 是面向复杂编码、计算机操作和专业工作流的高性能模型，具备接近 GPT‑6 Astra 的能力，同时响应更快、成本更低。支持文本和图像输入、文本输出、推理、函数调用、结构化输出及工具调用。","icon":"OpenAI","tags":"对话,思考,联网思考","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":2,"enable_groups":["default","GPT","GPT_K","OpenAI","wy_vip"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 272000 ? tier(\"short context\", p * 2 + c * 10 + cr * 0.1 + cc * 2.5) : p \u003e 272000 ? tier(\"long context\", p * 4 + c * 15 + cr * 0.2 + cc * 5) : tier(\"default\", p * 4 + c * 15 + cr * 0.2 + cc * 5)"},{"model_name":"gpt-realtime-2","icon":"OpenAI","tags":"实时语音","vendor_id":6,"quota_type":0,"model_ratio":2,"model_price":0,"owner_by":"","completion_ratio":6,"image_ratio":1.25,"audio_ratio":8,"audio_completion_ratio":2,"enable_groups":["OpenAI","wy_vip","default","GPT","GPT_K"],"supported_endpoint_types":["openai","openai-realtime"]},{"model_name":"happyhorse-1.0-r2v","description":"HappyHorse-1.0-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。","icon":"Qwen.Color","tags":"文生视频","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","HappyHouse"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"claude-haiku-5-5","description":"专为高吞吐量、对延迟敏感的工作而构建，例如分类、路由、提取和子智能体任务。它支持带有 effort 参数的 adaptive thinking（自适应思考）、1M 令牌的上下文窗口，以及最多 128k 输出令牌。它使用与 Claude 4.7 及更高版本模型相同的较新 tokenizer（分词器），因此相同文本的令牌数比在 Claude Haiku 4.5 上多约 30%。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["CCAnthropic","default","Anthropic","Anthropic_U","AwsAnthropic","bingyu_vip"],"supported_endpoint_types":["anthropic","openai"],"billing_mode":"tiered_expr","billing_expr":"len \u003c= 100000 ? tier(\"short content\", p * 0.1 + c * 0.5 + cr * 0.01 + cc * 0.125 + cc1h * 0.2) : len \u003e 100000 ? tier(\"long content\", p * 0.5 + c * 2.5 + cr * 0.05 + cc * 0.625 + cc1h * 1) : tier(\"default\", p * 0.5 + c * 2.5 + cr * 0.05 + cc * 0.625 + cc1h * 1)"},{"model_name":"claude-sonnet-5","description":"Claude Sonnet 第五代主力模型，在 Sonnet 4.6 基础上全面升级编码、工具调用与长程 Agent 能力，兼顾性能与成本，是日常开发与 Agent 工作流的首选平衡型旗舰。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":1,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["default","Anthropic","Anthropic_U","AwsAnthropic","bingyu_vip","CCAnthropic"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"seedance-2-0-mini","description":"Doubao-Seedance-2.0-mini 是豆包大模型团队推出的高性价比视频生成模型，面向更广泛的视频创作与生产需求\n相较 Seedance 2.0，2.0 mini 在价格上进一步下探，生成成本降低约 50%，帮助企业客户以更低成本完成调用。2.0 mini 兼顾成本与可用性，适用于电商内容生产、营销素材批量生成、UGC 内容创作、特效玩法生成等高频、规模化的视频生成场景","icon":"ByteDance.Color","tags":"文生视频,图生视频","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["Seedance","Seedance_1","bingyu_vip"],"supported_endpoint_types":["openai","video","video-generation"]},{"model_name":"dola-seedream-5-0-pro-260628","description":"BytePlus Seedream 5.0 Pro 图片生成模型，支持文生图和图生图，适用于高质量图片创作、图片编辑和多参考图生成。","icon":"ByteDance.Color","tags":"文生图,图生图","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["Seedream5Pro","bingyu_vip","BytePlus","default","Seedream"],"supported_endpoint_types":["openai","image-generation"]},{"model_name":"gemini-3.1-flash-image","icon":"Gemini.Color","tags":"文生图,图生图","vendor_id":3,"quota_type":0,"model_ratio":0.25,"model_price":0,"owner_by":"","completion_ratio":6,"cache_ratio":0.1,"image_ratio":1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai","image-generation"]},{"model_name":"seedance-2-0-mini-NSFW","description":"Seedance 2.0 mini是一款新一代的高性价比视频生成模型，专为满足更广泛的视频创作需求而设计。","icon":"ByteDance.Color","tags":"文生视频,图生视频","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["Seedance","Seedance_1","bingyu_vip"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"wan2.7-videoedit","description":"Wan2.7-VideoEdit，自然语言指令编辑视频，支持局部或全局编辑，可参考图像替换视频元素，支持复刻视频动作、特效、运镜等动态过程。","icon":"Qwen.Color","tags":"参考生视频","vendor_id":11,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","default","Wan"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"gpt-5.1-chat","description":"GPT-5.1-chat：首次在聊天场景引入链式思维，带来更快、更自然、更具上下文感知的对话体验。它不仅响应快，还强化安全和指令遵循，非常适用于客服、IT Helpdesk、HR、销售支持等场景。支持多模态聊天（文本、图像、音频），让长轮次对话也能保持一致性，助力企业实现品牌一致、安全合规的智能沟通。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.625,"model_price":0,"owner_by":"","completion_ratio":8,"cache_ratio":0.104,"enable_groups":["default","OpenAI","unal_vip","xy_vip"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"gpt-5.4-mini","description":"GPT‑5.4  是 OpenAI 最新推出、面向专业工作场景的高性能模型，集成了更强的推理、编程与智能体能力，能更精准高效地完成复杂任务，减少反复沟通。它支持可查看、可中途调整的思考过程，搭配更强的网页搜索与长上下文理解，同时原生具备计算机操作能力，上下文长度可达 1M Token，推理更省 Token、响应更快，适合高强度、高复杂度的专业工作。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.375,"model_price":0,"owner_by":"","completion_ratio":6,"cache_ratio":0.1,"enable_groups":["xy_vip","AzureGPT","bingyu_vip","default","OpenAI","unal_vip"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"gpt-5.4","description":"GPT‑5.4 是 OpenAI 最新推出、面向专业工作场景的高性能模型，集成了更强的推理、编程与智能体能力，能更精准高效地完成复杂任务，减少反复沟通。它支持可查看、可中途调整的思考过程，搭配更强的网页搜索与长上下文理解，同时原生具备计算机操作能力，上下文长度可达 1M Token，推理更省 Token、响应更快，适合高强度、高复杂度的专业工作。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":6,"enable_groups":["unal_vip","wy_vip","xy_vip","default","GPT","GPT_K","OpenAI"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 272000 ? tier(\"short context\", p * 2.5 + c * 15 + cr * 0.25) : p \u003e 272000 ? tier(\"long context\", p * 5 + c * 22.5 + cr * 0.5) : tier(\"default\", p * 5 + c * 22.5 + cr * 0.5)"},{"model_name":"claude-opus-4-5-20251101","description":"Claude Opus 4.5 是 Anthropic 最智能的模型，为生产代码、复杂代理和复杂办公任务设定了新标准。Opus 4.5 提供最先进的编码能力，性能比之前的 Opus 级智能降低了三分之一。它在需要从多个数据源获取信息、处理模糊性并无需人工指导进行权衡推理的复杂、长时间运行的任务中表现优异。","icon":"Claude.Color","tags":"对话,思考","vendor_id":2,"quota_type":0,"model_ratio":2.5,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"create_cache_ratio":1.25,"enable_groups":["bingyu_vip","CCAnthropic","default","Anthropic","Anthropic_U","AwsAnthropic"],"supported_endpoint_types":["anthropic","openai"]},{"model_name":"gemini-omni-flash-preview","description":"Gemini Omni Flash（预览版）是一种多模态模型，专为视频、图片和文本任务而设计。它针对视频生成进行了优化，可在单个模型中提供视频输出以及文本响应。","icon":"Gemini.Color","tags":"文生视频,图生视频","vendor_id":3,"quota_type":0,"model_ratio":0.75,"model_price":0,"owner_by":"","completion_ratio":6,"audio_ratio":1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai","openai-video","google-interactions"]},{"model_name":"seed-2-0-lite-260428","icon":"ByteDance.Color","tags":"对话,思考,联网搜索","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["BytePlus","default","vip","bingyu_vip"],"supported_endpoint_types":["openai","anthropic","openai-response"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 128000 ? tier(\"short context\", p * 0.25 + c * 2 + cr * 0.05 + cc * 0.008333 + ai * 3.75) : p \u003e 128000 \u0026\u0026 p \u003c= 256000 ? tier(\"long context\", p * 0.5 + c * 4 + cr * 0.1 + cc * 0.008333 + ai * 7.5) : tier(\"default\", p * 0.5 + c * 4 + cr * 0.1 + cc * 0.008333 + ai * 7.5)"},{"model_name":"seedance-2-0-fast","description":"Seedance 2.0 fast 作为全球领先的视频生成模型，可生成音画高精同步的视频内容。支持多人多语言对白，全面覆盖环境音、动作音、合成音、乐器音、背景音及人声，支持首尾帧，实现影视级叙事效果，满足影视、漫剧、电商及广告领域的高阶创作需求。","icon":"ByteDance.Color","tags":"文生视频","vendor_id":4,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["bingyu_vip","Seedance","Seedance_1"],"supported_endpoint_types":["openai","video-generation"]},{"model_name":"glm-5.1","description":"GLM-5.1是智谱AI推出的面向长程任务（Long Horizon Task）设计的模型，总参数744B，支持200K超长上下文，最大输出 128K tokens。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。","icon":"Zhipu.Color","tags":"对话,思考","vendor_id":23,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","GLM","GLM_Dep"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"],"billing_mode":"tiered_expr","billing_expr":"len \u003c= 32000 ? tier(\"short context\", p * 0.8823 + c * 3.5294 + cr * 0.1911) : len \u003e 32000 ? tier(\"long context\", p * 1.1764 + c * 4.1176 + cr * 0.2941) : tier(\"第3档\", p * 1.1764 + c * 4.1176 + cr * 0.2941)"},{"model_name":"gpt-5.3-codex","description":"gpt-5.3-codex 专为可操控性、前端开发和交互性而设计。\n关键功能：\n1）多模态推理：接受文本和图像输入——适用于在单一工作流中推理代码、UI 状态、架构图和设计草图。\n2）基于上下文的代码审查：通过 AI 驱动的洞察力，揭示关键流程和质量问题，针对您的开发界面。\n3）异步执行：支持长时间运行的任务，通过模型引导的循环保留状态和上下文，跨交互进行。\n4）仓库感知智能：理解完整仓库，支持连贯的重构、测试自动化以及通过截图进行可视化验证。","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.875,"model_price":0,"owner_by":"","completion_ratio":8,"cache_ratio":0.1,"enable_groups":["default","OpenAI","unal_vip","xy_vip"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"qwen3.8-max","description":"2.4万亿参数MoE旗舰，编程与办公能力全面跃升，可自主编程十数天交付完整项目。胜任法律、金融、设计等数百种专业任务，一次对话端到端交付生产级成果。原生视觉理解贯穿规划、执行与验证全流程，支持超长文档与长视频的深度语义解析。长程任务中自主规划与闭环迭代，持续进化。","icon":"Qwen.Color","tags":"对话,思考","vendor_id":11,"quota_type":0,"model_ratio":0.88235,"model_price":0,"owner_by":"","completion_ratio":3,"cache_ratio":0.125007083357,"enable_groups":["default","Qwen","vip"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"text-embedding-ada-002","description":"text-embedding-ada-002 在文本搜索、代码搜索和句子相似性任务上优于所有早期的嵌入模型，并在文本分类任务上获得相当的性能。\n关键模型功能：\n1）文本搜索\n2）代码搜索\n3）句子相似度任务\n4）文本分类\n注意：此模型可用于推理，特别是用于嵌入，但不能进行微调。","icon":"OpenAI","tags":"嵌入","vendor_id":6,"quota_type":0,"model_ratio":0.05,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","GPT","GPT_K","OpenAI","wy_vip"],"supported_endpoint_types":["openai","embeddings"]},{"model_name":"gemini-2.5-flash","description":"Gemini-2.5-Flash 是一款超快速的 AI 模型，就像一个闪电侠。它能在瞬间处理信息，特别适合需要快速反应的任务，比如实时对话或者快速生成内容。和其他模型相比，它速度快、效率高，还能处理多种格式的内容，是个超级高效的“小助手”。\n输入：文本，图片，视频，音频\n输出：文本","icon":"Gemini.Color","tags":"对话,思考","vendor_id":3,"quota_type":0,"model_ratio":0.15,"model_price":0,"owner_by":"","completion_ratio":8.333333333333,"cache_ratio":0.1,"audio_ratio":3.333333333333,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"gemini-3.1-pro-preview","description":"Gemini 3.1 Pro Preview 具备强大的编码能力和最先进的推理能力，以及复杂的多模态理解能力。与 Gemini 2.5 Pro 相比，它在复杂指令遵循方面有显著提升，并提供了更好的输出效率。它具有 1M 令牌的上下文窗口，并具备复杂的多模态理解能力。","icon":"Gemini.Color","tags":"对话,思考,联网搜索","vendor_id":3,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":4,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"],"billing_mode":"tiered_expr","billing_expr":"p \u003c= 200000 ? tier(\"short context\", p * 2 + c * 12 + cr * 0.2) : p \u003e 200000 ? tier(\"long context\", p * 4 + c * 18 + cr * 0.4) : tier(\"default\", p * 2 + c * 12 + cr * 0.2)"},{"model_name":"gemini-3.6-flash","description":"Gemini 3.6 Flash 针对多步编排、全栈代码重构和一般推理进行了优化。它在对 Flash 系列模型至关重要的许多关键领域进行了改进。","icon":"Gemini.Color","tags":"对话,思考,联网搜索","vendor_id":3,"quota_type":0,"model_ratio":0.375,"model_price":0,"owner_by":"","completion_ratio":5,"cache_ratio":0.1,"enable_groups":["bingyu_vip","default","Google","Vertex"],"supported_endpoint_types":["gemini","openai"]},{"model_name":"qwen3.7-max","icon":"Qwen.Color","tags":"对话,思考,联网搜索","vendor_id":11,"quota_type":0,"model_ratio":0.88235,"model_price":0,"owner_by":"","completion_ratio":3,"cache_ratio":0.199977333258,"enable_groups":["default","Qwen","vip","bingyu_vip"],"supported_endpoint_types":["openai","openai-response","openai-response-compact","anthropic","gemini","openai-alpha-search"]},{"model_name":"gpt-5.1","description":"GPT-5.1 重点提升“按需推理”能力，面对不同难度的任务可自动调整推理深度。核心能力包括：\n1）自适应推理：根据上下文灵活调整思考时间\n2）多模态智能：支持文本、图片、音频的输入输出\n3）企业级性能：安全与合规","icon":"OpenAI","tags":"对话,思考,联网搜索","vendor_id":6,"quota_type":0,"model_ratio":0.625,"model_price":0,"owner_by":"","completion_ratio":8,"cache_ratio":0.1,"enable_groups":["xy_vip","default","OpenAI","unal_vip"],"supported_endpoint_types":["openai","openai-response"]},{"model_name":"grok-4.7","description":"Grok 4.7 是 xAI 新一代旗舰大模型，专注于编程、智能体任务和专业知识工作。模型采用更大规模的基础模型与更长的强化学习训练，在复杂问题推理、代码开发、多步骤 Agent 工作流、文档与专业任务处理方面进一步提升，并具备更强的自我验证和长上下文管理能力。支持文本与图片输入，支持函数调用、结构化输出、推理以及 Web Search、X Search 和代码执行等工具能力，适用于软件开发、复杂分析、自动化 Agent、专业办公等场景。","icon":"Grok","tags":"对话,思考","vendor_id":14,"quota_type":0,"model_ratio":37.5,"model_price":0,"owner_by":"","completion_ratio":1,"enable_groups":["default","Grok"],"supported_endpoint_types":["openai","openai-response"],"billing_mode":"tiered_expr","billing_expr":"len \u003c= 200000 ? tier(\"short context\", p * 2 + c * 6 + cr * 0.3) : len \u003e 200000 ? tier(\"long context\", p * 4 + c * 12 + cr * 1) : tier(\"第3档\", p * 4 + c * 12 + cr * 1)"}],"group_ratio":{"Seedance":1.05,"Seedance_1":1.05,"Seedance_Old":1,"auto":1,"default":1},"pricing_version":"a42d372ccf0b5dd13ecf71203521f9d2","success":true,"supported_endpoint":{"anthropic":{"path":"/v1/messages","method":"POST"},"embeddings":{"path":"{     \"path\": \"/v1/embeddings\",     \"method\": \"POST\" }","method":"POST"},"gemini":{"path":"/v1beta/models/{model}:generateContent","method":"POST"},"google-interactions":{"path":"/v1beta/interactions","method":"POST"},"image-generation":{"path":"/v1/images/generations","method":"POST"},"images-edits":{"path":"/v1/images/edits","method":"POST"},"openai":{"path":"/v1/chat/completions","method":"POST"},"openai-alpha-search":{"path":"/v1/alpha/search","method":"POST"},"openai-images":{"path":"/v1/images/generations","method":"POST"},"openai-images-edits":{"path":"/v1/images/edits","method":"POST"},"openai-realtime":{"path":"/v1/realtime","method":"POST"},"openai-response":{"path":"/v1/responses","method":"POST"},"openai-response-compact":{"path":"/v1/responses/compact","method":"POST"},"openai-speech":{"path":"/v1/audio/speech","method":"POST"},"openai-video":{"path":"/v1/videos","method":"POST"},"video":{"path":"/v1/video/generations","method":"POST"},"video-generation":{"path":"/api/v3/contents/generations/tasks","method":"POST"}},"usable_group":{"Seedance":"Seedance默认分组","Seedance_1":"Seedance新分组","Seedance_Old":"Seedance","auto":"自动","default":"默认分组"},"vendors":[{"id":37,"name":"阿里巴巴","icon":"Qwen.Color"},{"id":14,"name":"xAI","icon":"XAI"},{"id":2,"name":"Anthropic","icon":"Claude.Color"},{"id":23,"name":"Zhipu","description":"Zhipu.Color","icon":"Zhipu.Color"},{"id":32,"name":"Kling","icon":"Kling.Color"},{"id":39,"name":"Hunyuan","icon":"Hunyuan.Color"},{"id":3,"name":"Google","icon":"Gemini.Color"},{"id":6,"name":"OpenAI","icon":"OpenAI"},{"id":7,"name":"JiMeng","icon":"Jimeng.Color"},{"id":16,"name":"Meta","icon":"Ollama"},{"id":31,"name":"字节跳动","icon":"Doubao.Color"},{"id":36,"name":"Xiaomi","icon":"XiaomiMiMo"},{"id":38,"name":"智谱","icon":"Zhipu.Color"},{"id":4,"name":"BytePlus","description":"BytePlus","icon":"ByteDance.Color"},{"id":5,"name":"DeepSeek","description":"DeepSeek","icon":"DeepSeek.Color"},{"id":11,"name":"Ali","description":"Tongyi","icon":"Qwen.Color"},{"id":19,"name":"Moonshot","icon":"Moonshot"},{"id":29,"name":"MiniMax","description":"Minimax","icon":"Minimax.Color"}]}