Models & reference pricing

Claude Fable 5 is Anthropic's highest-capability Claude tier for demanding reasoning, complex coding, and long-running agent workflows. It provides a 1M-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching, PDF input. Use model ID anthropic/claude-fable-5 through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

Claude Haiku 4.5 is a fast, cost-efficient Claude model for responsive automation, coding assistance, and high-volume everyday tasks. It provides a 200K-token context window and up to 64K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, prompt caching, PDF input. Use model ID anthropic/claude-haiku-4.5 through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

Claude Opus 4.5 is a high-capability Claude Opus model for complex analysis, software engineering, and multi-step agent workflows. It provides a 200K-token context window and up to 64K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, prompt caching, PDF input. Use model ID anthropic/claude-opus-4.5 through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

Anthropic Claude Opus 4.6 is a high-capability Opus model released on February 5, 2026, with a 1M-token context window and up to 128K output tokens for complex coding, long-running agent workflows, and demanding knowledge work. It supports vision, function calling, reasoning, prompt caching, and PDF input, and is available through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

Claude Opus 4.7 is a high-capability Claude Opus model for complex coding, web-connected research, and long-running agent workflows. It provides a 1M-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching, PDF input. Use model ID anthropic/claude-opus-4.7 through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

Claude Opus 4.8 is an advanced Claude Opus model for demanding reasoning, structured outputs, document work, and complex software engineering. It provides a 1M-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, prompt caching, PDF input, structured outputs. Use model ID anthropic/claude-opus-4.8 through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

Anthropic's flagship Claude model for demanding reasoning, advanced coding, visual analysis, and long-horizon agent workflows.

Claude Sonnet 4.5 is a balanced Claude Sonnet model for coding, analysis, document processing, and general agent workflows. It provides a 200K-token context window and up to 64K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, prompt caching, PDF input. Use model ID anthropic/claude-sonnet-4.5 through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

Claude Sonnet 4.6 is a balanced high-performance Claude Sonnet model for coding, reasoning, web-connected research, and production agents. It provides a 200K-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching, PDF input. Use model ID anthropic/claude-sonnet-4.6 through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

Claude Sonnet 5 is a next-generation Claude Sonnet model for advanced coding, reasoning, research, and scalable agent workflows. It provides a 1M-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching, PDF input. Use model ID anthropic/claude-sonnet-5 through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

SpaceXAI's frontier model for coding, agentic tasks, and knowledge work.

SpaceXAI's smartest model for frontier coding, knowledge work, and STEM tasks.

Text-only coding model for Grok Build workflows.

GPT-5.3 Codex 是 OpenAI 面向代理式编码的高能力模型，结合 GPT-5.2 Codex 的软件工程能力和更广泛的推理、专业知识能力，适合多语言编码、终端任务和真实计算机使用场景。

GPT-5.4 is a flagship GPT model for complex reasoning, coding, multimodal analysis, web research, and long-running agents. It provides a 272K-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching, code interpreter. Use model ID openai/gpt-5.4 through kineticRouter's OpenAI-compatible API.

GPT-5.4 Mini is a balanced GPT-5.4 tier for everyday coding, reasoning, multimodal tasks, and cost-sensitive agent workflows. It provides a 272K-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching, code interpreter. Use model ID openai/gpt-5.4-mini through kineticRouter's OpenAI-compatible API.

GPT-5.5 is an advanced GPT model for professional reasoning, coding, multimodal analysis, and agent workflows. It provides a 272K-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching, code interpreter. Use model ID openai/gpt-5.5 through kineticRouter's OpenAI-compatible API.

GPT-5.6 Luna is the fast, cost-efficient GPT-5.6 tier for high-volume, latency-sensitive chat, classification, and lightweight agents. It provides a 372K-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching. Use model ID openai/gpt-5.6-luna through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

GPT-5.6 Sol is the flagship GPT-5.6 tier for complex reasoning, coding, command-line tasks, and long-horizon agent workflows. It provides a 372K-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching. Use model ID openai/gpt-5.6-sol through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.

GPT-5.6 Terra is the balanced GPT-5.6 tier for everyday coding, reasoning, and agent workflows that need a strong capability-to-cost tradeoff. It provides a 372K-token context window and up to 128K output tokens. Its current kineticRouter catalog capabilities include vision, function calling, reasoning, web search, prompt caching. Use model ID openai/gpt-5.6-terra through kineticRouter's OpenAI-compatible and Anthropic-compatible APIs.