@lobehub/chat
Version:
Lobe Chat - an open-source, high-performance chatbot framework that supports speech synthesis, multimodal, and extensible Function Call plugin system. Supports one-click free deployment of your private ChatGPT/LLM web application.
608 lines (605 loc) • 16.9 kB
text/typescript
import { AIChatModelCard } from '@/types/aiModel';
const aihubmixModels: AIChatModelCard[] = [
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'o4-mini 是我们最新的小型 o 系列模型。 它专为快速有效的推理而优化，在编码和视觉任务中表现出极高的效率和性能。',
displayName: 'o4-mini',
enabled: true,
id: 'o4-mini',
maxOutput: 100_000,
pricing: {
cachedInput: 0.275,
input: 1.1,
output: 4.4,
},
releasedAt: '2025-04-17',
settings: {
extendParams: ['reasoningEffort'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'o4-mini-deep-research 是我们更快速、更实惠的深度研究模型——非常适合处理复杂的多步骤研究任务。它可以从互联网上搜索和综合信息，也可以通过 MCP 连接器访问并利用你的自有数据。',
displayName: 'o4-mini Deep Research',
id: 'o4-mini-deep-research',
maxOutput: 100_000,
pricing: {
cachedInput: 0.5,
input: 2,
output: 8,
},
releasedAt: '2025-06-26',
settings: {
extendParams: ['reasoningEffort'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'o3-pro 模型使用更多的计算来更深入地思考并始终提供更好的答案，仅支持 Responses API 下使用。',
displayName: 'o3-pro',
id: 'o3-pro',
maxOutput: 100_000,
pricing: {
input: 20,
output: 80,
},
releasedAt: '2025-06-10',
settings: {
extendParams: ['reasoningEffort'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'o3 是一款全能强大的模型，在多个领域表现出色。它为数学、科学、编程和视觉推理任务树立了新标杆。它也擅长技术写作和指令遵循。用户可利用它分析文本、代码和图像，解决多步骤的复杂问题。',
displayName: 'o3',
enabled: true,
id: 'o3',
maxOutput: 100_000,
pricing: {
cachedInput: 0.5,
input: 2,
output: 8,
},
releasedAt: '2025-04-16',
settings: {
extendParams: ['reasoningEffort'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'o3-deep-research 是我们最先进的深度研究模型，专为处理复杂的多步骤研究任务而设计。它可以从互联网上搜索和综合信息，也可以通过 MCP 连接器访问并利用你的自有数据。',
displayName: 'o3 Deep Research',
id: 'o3-deep-research',
maxOutput: 100_000,
pricing: {
cachedInput: 2.5,
input: 10,
output: 40,
},
releasedAt: '2025-06-26',
settings: {
extendParams: ['reasoningEffort'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
search: true,
vision: true,
},
contextWindowTokens: 1_047_576,
description: 'GPT-4.1 是我们用于复杂任务的旗舰模型。它非常适合跨领域解决问题。',
displayName: 'GPT-4.1',
enabled: true,
id: 'gpt-4.1',
maxOutput: 32_768,
pricing: {
cachedInput: 0.5,
input: 2,
output: 8,
},
releasedAt: '2025-04-14',
settings: {
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
search: true,
vision: true,
},
contextWindowTokens: 1_047_576,
description:
'GPT-4.1 mini 提供了智能、速度和成本之间的平衡，使其成为许多用例中有吸引力的模型。',
displayName: 'GPT-4.1 mini',
id: 'gpt-4.1-mini',
maxOutput: 32_768,
pricing: {
cachedInput: 0.1,
input: 0.4,
output: 1.6,
},
releasedAt: '2025-04-14',
settings: {
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
vision: true,
},
contextWindowTokens: 1_047_576,
description: 'GPT-4.1 nano 是最快，最具成本效益的GPT-4.1模型。',
displayName: 'GPT-4.1 nano',
id: 'gpt-4.1-nano',
maxOutput: 32_768,
pricing: {
cachedInput: 0.025,
input: 0.1,
output: 0.4,
},
releasedAt: '2025-04-14',
type: 'chat',
},
{
abilities: {
vision: true,
},
contextWindowTokens: 128_000,
description:
'ChatGPT-4o 是一款动态模型，实时更新以保持当前最新版本。它结合了强大的语言理解与生成能力，适合于大规模应用场景，包括客户服务、教育和技术支持。',
displayName: 'ChatGPT-4o',
enabled: true,
id: 'chatgpt-4o-latest',
pricing: {
input: 5,
output: 15,
},
type: 'chat',
},
{
abilities: {
functionCall: true,
search: true,
},
contextWindowTokens: 131_072,
description:
'旗舰级模型，擅长数据提取、编程和文本摘要等企业级应用，拥有金融、医疗、法律和科学等领域的深厚知识。',
displayName: 'Grok 3',
enabled: true,
id: 'grok-3',
pricing: {
cachedInput: 0.75,
input: 3,
output: 15,
},
releasedAt: '2025-04-03',
settings: {
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
},
contextWindowTokens: 131_072,
description:
'轻量级模型，回话前会先思考。运行快速、智能，适用于不需要深层领域知识的逻辑任务，并能获取原始的思维轨迹。',
displayName: 'Grok 3 Mini',
enabled: true,
id: 'grok-3-mini',
pricing: {
input: 0.3,
output: 0.5,
},
releasedAt: '2025-04-03',
settings: {
extendParams: ['reasoningEffort'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'Claude Opus 4.1 是 Anthropic 最新的用于处理高度复杂任务的最强大模型。它在性能、智能、流畅性和理解力方面表现卓越。',
displayName: 'Claude Opus 4.1',
enabled: true,
id: 'claude-opus-4-1-20250805',
maxOutput: 32_000,
pricing: {
cachedInput: 1.5,
input: 16.5,
output: 82.5,
writeCacheInput: 18.75,
},
releasedAt: '2025-08-05',
settings: {
extendParams: ['disableContextCaching', 'enableReasoning', 'reasoningBudgetToken'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'Claude Opus 4 是 Anthropic 用于处理高度复杂任务的最强大模型。它在性能、智能、流畅性和理解力方面表现卓越。',
displayName: 'Claude Opus 4',
id: 'claude-opus-4-20250514',
maxOutput: 32_000,
pricing: {
cachedInput: 1.5,
input: 16.8,
output: 84,
writeCacheInput: 18.75,
},
releasedAt: '2025-05-23',
settings: {
extendParams: ['disableContextCaching', 'enableReasoning', 'reasoningBudgetToken'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'Claude Sonnet 4 可以产生近乎即时的响应或延长的逐步思考，用户可以清晰地看到这些过程。API 用户还可以对模型思考的时间进行细致的控制',
displayName: 'Claude Sonnet 4',
enabled: true,
id: 'claude-sonnet-4-20250514',
maxOutput: 64_000,
pricing: {
cachedInput: 0.3,
input: 3.3,
output: 16.5,
writeCacheInput: 3.75,
},
releasedAt: '2025-05-23',
settings: {
extendParams: ['disableContextCaching', 'enableReasoning', 'reasoningBudgetToken'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'Claude 3.7 Sonnet 是 Anthropic 迄今为止最智能的模型，也是市场上首个混合推理模型。Claude 3.7 Sonnet 可以产生近乎即时的响应或延长的逐步思考，用户可以清晰地看到这些过程。Sonnet 特别擅长编程、数据科学、视觉处理、代理任务。',
displayName: 'Claude 3.7 Sonnet',
id: 'claude-3-7-sonnet-20250219',
maxOutput: 64_000,
pricing: {
cachedInput: 0.3,
input: 3.3,
output: 16.5,
writeCacheInput: 3.75,
},
releasedAt: '2025-02-24',
settings: {
extendParams: ['disableContextCaching', 'enableReasoning', 'reasoningBudgetToken'],
searchImpl: 'params',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
vision: true,
},
contextWindowTokens: 200_000,
description:
'Claude 3.5 Haiku 是 Anthropic 最快的下一代模型。与 Claude 3 Haiku 相比，Claude 3.5 Haiku 在各项技能上都有所提升，并在许多智力基准测试中超越了上一代最大的模型 Claude 3 Opus。',
displayName: 'Claude 3.5 Haiku',
id: 'claude-3-5-haiku-20241022',
maxOutput: 8192,
pricing: {
cachedInput: 0.1,
input: 1.1,
output: 5.5,
writeCacheInput: 1.25,
},
releasedAt: '2024-11-05',
settings: {
extendParams: ['disableContextCaching'],
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
},
contextWindowTokens: 65_536,
description:
'已升级至最新版本250528；字节火山云开源部署的满血 R1，总参数量 671B，输入最高 64k。目前最稳定，推荐用这个。',
displayName: 'DeepSeek R1',
enabled: true,
id: 'DeepSeek-R1',
pricing: {
input: 0.546,
output: 2.184,
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
},
contextWindowTokens: 65_536,
description:
'由微软部署提供； DeepSeek R1型号已进行小版本升级，当前版本为DeepSeek-R1-0528。在最新的更新中，DeepSeek R1通过增加计算资源和引入后训练阶段的算法优化机制，大幅提升了推理深度和推断能力。该模型在数学、编程和通用逻辑等多个基准测试中表现出色，其整体性能已接近领先模型，如O3和Gemini 2.5 Pro 。',
displayName: 'DeepSeek R1 0528 (Azure)',
id: 'azure-DeepSeek-R1-0528',
pricing: {
input: 0.4,
output: 1.6,
},
type: 'chat',
},
{
abilities: {
functionCall: true,
},
contextWindowTokens: 65_536,
description: '字节火山云开源部署目前最稳定，推荐用这个。已经自动升级为最新发布的版本 250324 。',
displayName: 'DeepSeek V3',
enabled: true,
id: 'DeepSeek-V3',
pricing: {
input: 0.272,
output: 1.088,
},
type: 'chat',
},
{
abilities: {
functionCall: true,
},
contextWindowTokens: 65_536,
description:
'模型供应商为：sophnet平台。DeepSeek V3 Fast 是 DeepSeek V3 0324 版本的高TPS极速版，满血非量化，代码与数学能力更强，响应更快！',
displayName: 'DeepSeek V3 (Fast)',
id: 'DeepSeek-V3-Fast',
pricing: {
input: 0.55,
output: 2.2,
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 1_048_576 + 65_536,
description:
'Gemini 2.5 Pro 是 Google 最先进的思维模型，能够对代码、数学和STEM领域的复杂问题进行推理，以及使用长上下文分析大型数据集、代码库和文档。',
displayName: 'Gemini 2.5 Pro',
enabled: true,
id: 'gemini-2.5-pro',
maxOutput: 65_536,
pricing: {
input: 1.25, // prompts <= 200k tokens
output: 10, // prompts <= 200k tokens
},
releasedAt: '2025-06-17',
settings: {
extendParams: ['thinkingBudget'],
searchImpl: 'params',
searchProvider: 'google',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 1_048_576 + 65_536,
description: 'Gemini 2.5 Flash 是 Google 性价比最高的模型，提供全面的功能。',
displayName: 'Gemini 2.5 Flash',
enabled: true,
id: 'gemini-2.5-flash',
maxOutput: 65_536,
pricing: {
cachedInput: 0.075,
input: 0.3,
output: 2.5,
},
releasedAt: '2025-06-17',
settings: {
extendParams: ['thinkingBudget'],
searchImpl: 'params',
searchProvider: 'google',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
search: true,
vision: true,
},
contextWindowTokens: 1_048_576 + 65_536,
description: 'Gemini 2.5 Flash-Lite 是 Google 最小、性价比最高的模型，专为大规模使用而设计。',
displayName: 'Gemini 2.5 Flash-Lite',
id: 'gemini-2.5-flash-lite',
maxOutput: 65_536,
pricing: {
cachedInput: 0.025,
input: 0.1,
output: 0.4,
},
releasedAt: '2025-07-22',
settings: {
extendParams: ['thinkingBudget'],
searchImpl: 'params',
searchProvider: 'google',
},
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
},
contextWindowTokens: 131_072,
description:
'基于Qwen3的思考模式开源模型，相较上一版本（通义千问3-235B-A22B）逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景。',
displayName: 'Qwen3 235B A22B Thinking 2507',
enabled: true,
id: 'qwen3-235b-a22b-thinking-2507',
maxOutput: 32_768,
organization: 'Qwen',
pricing: {
currency: 'CNY',
input: 0.28,
output: 2.8,
},
releasedAt: '2025-07-25',
type: 'chat',
},
{
abilities: {
functionCall: true,
},
contextWindowTokens: 131_072,
description:
'基于Qwen3的非思考模式开源模型，相较上一版本（通义千问3-235B-A22B）主观创作能力与模型安全性均有小幅度提升。',
displayName: 'Qwen3 235B A22B Instruct 2507',
enabled: true,
id: 'qwen3-235b-a22b-instruct-2507',
maxOutput: 32_768,
organization: 'Qwen',
pricing: {
currency: 'CNY',
input: 0.28,
output: 1.12,
},
releasedAt: '2025-07-22',
type: 'chat',
},
{
abilities: {
functionCall: true,
reasoning: true,
},
contextWindowTokens: 131_072,
description:
'基于Qwen3的思考模式开源模型，相较上一版本（通义千问3-30B-A3B）逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景。',
displayName: 'Qwen3 30B A3B Thinking 2507',
enabled: true,
id: 'qwen3-30b-a3b-thinking-2507',
maxOutput: 32_768,
organization: 'Qwen',
pricing: {
currency: 'CNY',
input: 0.12,
output: 1.2,
},
releasedAt: '2025-07-30',
type: 'chat',
},
{
abilities: {
functionCall: true,
},
contextWindowTokens: 131_072,
description:
'相较上一版本（Qwen3-30B-A3B）中英文和多语言整体通用能力有大幅提升。主观开放类任务专项优化，显著更加符合用户偏好，能够提供更有帮助性的回复。',
displayName: 'Qwen3 30B A3B Instruct 2507',
enabled: true,
id: 'qwen3-30b-a3b-instruct-2507',
maxOutput: 32_768,
organization: 'Qwen',
pricing: {
input: 0.12,
output: 0.48,
},
releasedAt: '2025-07-29',
type: 'chat',
},
];
export default aihubmixModels;