API Key — obtain from https://console.anthropic.com/

Claude 4 family — best reasoning, coding, and agentic capabilities

Anthropic

Claude 4 系列 — 最强推理、编程和 Agent 能力

Anthropic Claude

Most intelligent model for agents and coding

Claude Opus 4.6

Best combination of speed and intelligence

Claude Sonnet 4.6

API Key — obtain from https://platform.openai.com/signup

GPT-5 series — versatile AI with broad capabilities and 1M context

OpenAI

GPT-5 系列 — 功能全面的通用 AI，支持百万级上下文

GPT-5.4

Strongest mini model for coding, computer use, and subagents

GPT-5.4 mini

Codex specialized model for code generation

GPT-5.3 Codex

API Key — obtain from https://aistudio.google.com/

Gemini models — Google's latest AI with multimodal support and 1M context

Google

Gemini 系列 — Google 最新多模态 AI，支持百万级上下文

Google Gemini

Advanced intelligence with complex problem-solving, 1M context

Gemini 3.1 Pro Preview

Frontier-class performance for agentic workflows with configurable reasoning

Gemini 3 Flash Preview

Most advanced stable model for complex tasks

Gemini 2.5 Pro

Best price-performance for low-latency, high-volume tasks

Gemini 2.5 Flash

European AI lab. Strong coding, reasoning, and agent capabilities with competitive pricing.

Mistral

Flagship open-weight multimodal model for general-purpose tasks

Mistral Large 3

Hybrid model combining instruct, reasoning, and coding. Best value.

Mistral Small 4

Frontier code agent model for software engineering tasks. Built for tool use and codebase exploration.

Devstral 2

API Key — obtain from https://openrouter.ai/keys

Aggregator with 200+ models. Often offers lower prices for open-source models via GPU cloud providers.

OpenRouter

聚合平台 — 通过单一 OpenAI 兼容 API 访问 200+ AI 模型

DeepSeek V3.2 (via OpenRouter)

DeepSeek R1 (via OpenRouter)

Kimi K2.5 (via OpenRouter)

GLM 4.7 (via OpenRouter)

GLM 5 (via OpenRouter)

Qwen 3 Max (via OpenRouter)

MiniMax M2.5 (via OpenRouter)

Alibaba Qwen 3.5 open-source 397B MoE model

Qwen 3.5 (via OpenRouter)

API Key — obtain from https://console.x.ai/

Grok models — fast inference with large 2M context and low hallucination

Grok 系列 — 超快推理、200万上下文、低幻觉率

xAI Grok

xAI's newest flagship model with reasoning, 2M context, image understanding

Grok 4.20 Beta (Reasoning)

xAI's newest flagship model without reasoning, 2M context, functions and structured outputs

Grok 4.20 Beta (Non-Reasoning)

Fast and affordable Grok model with reasoning, 2M context

Grok 4-1 Fast (Reasoning)

Fast and affordable Grok model without reasoning, 2M context

Grok 4-1 Fast (Non-Reasoning)

API Key — obtain from https://platform.deepseek.com/

DeepSeek V3.2 — unified model for chat and reasoning at ultra-competitive pricing

DeepSeek

DeepSeek V3.2 — 统一的对话和推理模型，极具竞争力的价格

DeepSeek (深度求索)

DeepSeek V3.2 non-thinking mode — general tasks, tool use, 8K max output

DeepSeek V3.2 (Chat)

DeepSeek V3.2 thinking mode — Chain-of-Thought reasoning, 64K max output

DeepSeek V3.2 (Reasoner)

API Key — obtain from https://dashscope.console.aliyun.com/

Qwen3 series — latest generation models with competitive pricing in China

Alibaba Cloud

Qwen3 系列 — 最新一代通义千问，国内高性价比

阿里云百炼 (通义千问)

Most capable Qwen3 model with deep thinking mode

Qwen3 Max

Balanced model with 1M context, supports text, image and video

Qwen3.5 Plus

API Key — obtain from https://platform.minimaxi.com/

MiniMax M2.5 — strong coding and agentic capabilities with competitive pricing

MiniMax

MiniMax M2.5 — 强编程和 Agent 能力，性价比高

MiniMax (稀宇科技)

MiniMax M2.7

SOTA model for productivity, coding (80.2% SWE-Bench), and agent workflows

MiniMax M2.5

API Key — obtain from https://platform.moonshot.cn/

Kimi K2.5 — strong reasoning and long context, China mainland API

Moonshot (China)

Kimi K2.5 — 强推理与长上下文，中国大陆 API

Moonshot / Kimi（中国站）

Flagship model with strong coding, agentic capabilities and vision

Kimi K2.5

API Key — obtain from https://platform.moonshot.ai/

Kimi K2.5 — strong reasoning and long context, international API

Moonshot (International)

Kimi K2.5 — 强推理与长上下文，国际 API

Moonshot / Kimi（国际站）

API Key — obtain from https://console.volcengine.com/

Doubao Seed and multi-model aggregation — ByteDance's cloud AI platform

Volcengine

豆包 Seed 及多模型聚合 — 字节跳动旗下云端 AI 平台

火山引擎 / 豆包

DeepSeek V3.2 hosted on Volcengine with 128K context

DeepSeek V3.2 (via Volcengine)

Moonshot's Kimi K2.5 hosted on Volcengine

Kimi K2.5 (via Volcengine)

API Key — obtain from https://cloud.baidu.com/

ERNIE 5.0 and third-party models — Baidu's enterprise AI platform

Baidu Qianfan

文心一言 ERNIE 5.0 及第三方模型 — 百度企业级 AI 平台

百度千帆

DeepSeek V3.2 hosted on Baidu Qianfan with 98K context

DeepSeek V3.2 (via Qianfan)

GLM-5 family — flagship reasoning and agentic models, China mainland API

Zhipu AI (China)

GLM-5 系列 — 旗舰推理与 Agent 模型，中国大陆 API

智谱 AI（中国站）

Optimized for agentic workflows, tool calling, and long-task stability

GLM-5-Turbo

针对 Agent 工作流、工具调用和长任务稳定性优化

New-generation flagship reasoning model for complex tasks

GLM-5

新一代旗舰推理模型，适合复杂任务

Enhanced programming, multi-step reasoning, and agent task performance

GLM-4.7

增强编程、多步推理和 Agent 任务性能

API Key — obtain from https://api.together.xyz/

Open-source model aggregator with precise per-model pricing and fast inference

Together AI

开源模型聚合平台，精确的按模型定价和快速推理

DeepSeek R1 reasoning model hosted on Together

DeepSeek R1

Moonshot's Kimi K2.5 hosted on Together, 262K context with vision

Kimi K2.5 (via Together)

Zhipu AI's GLM 4.7 hosted on Together, 202K context

GLM 4.7

Subscription access to Doubao, GLM, Kimi, MiniMax, DeepSeek via Volcengine Ark

Volcengine (Coding Plan)

订阅制，包含豆包、GLM、Kimi、MiniMax、DeepSeek 等模型

火山方舟（Coding Plan）

Doubao Seed 2.0 Pro

Doubao Seed 2.0 Code

DeepSeek V3.2

Subscription access to MiniMax, GLM, Kimi, DeepSeek via Baidu Qianfan

Baidu Qianfan (Coding Plan)

订阅制，包含 MiniMax、GLM、Kimi、DeepSeek 等模型

百度千帆（Coding Plan）

GLM 5

Subscription access to MiniMax, GLM, Kimi, Qwen via Alibaba Cloud Bailian

Alicloud (Coding Plan)

订阅制，包含 MiniMax、GLM、Kimi、Qwen 等模型

阿里云百炼（Coding Plan）

Qwen 3.5 Plus

Qwen 3 Max

Subscription access to MiniMax, GLM, Kimi, Hunyuan models via Tencent Cloud

Tencent Cloud (Coding Plan)

订阅制，包含 MiniMax、GLM、Kimi、混元 等模型

腾讯云（Coding Plan）

MiniMax original Coding Plan for China. MiniMax M2.7 with competitive pricing.

MiniMax Token Plan (China)

MiniMax 原厂编程订阅套餐，支持 M2.7 模型

MiniMax Token Plan（中国）

Latest MiniMax flagship model, ~50 TPS (100 TPS off-peak). Vision + web search + MCP.

MiniMax original Coding Plan for international users. MiniMax M2.7.

MiniMax Token Plan (International)

MiniMax international API. Same pricing as China for M2.5/M2.7.

MiniMax (International)

Latest MiniMax flagship. ~50 TPS, 100 TPS off-peak.

provider	models	regions	api type
Anthropic	2	us, eu	anthropic-messages
OpenAI	3	us, eu	openai-completions
Google	4	us, eu	google-generative-ai
Mistral	3	us, eu	openai-completions
OpenRouter	8	us	openai-completions
xAI	4	us	openai-completions
DeepSeek	2	cn, us	openai-completions
Alibaba Cloud	2	cn	openai-completions
MiniMax	2	cn	openai-completions
Moonshot (China)	1	cn	openai-completions
Moonshot (International)	1	us	openai-completions
Volcengine	2	cn	openai-completions
Baidu Qianfan	1	cn	openai-completions
Zhipu AI (China)	3	cn	openai-completions
Together AI	3	us	openai-completions
Volcengine (Coding Plan)	6	cn	openai-completions
Baidu Qianfan (Coding Plan)	5	cn	openai-completions
Alicloud (Coding Plan)	6	cn	openai-completions
Tencent Cloud (Coding Plan)	3	cn	openai-completions
MiniMax Token Plan (China)	1	cn	openai-completions
MiniMax Token Plan (International)	1	us, eu	openai-completions
MiniMax (International)	1	us, eu	openai-completions

models

by provider