LLM API for Every Leading Model
Call DeepSeek, Qwen, Doubao, GLM, Kimi and MiniMax models through one OpenAI-compatible API. Switch models by changing one parameter, and every model is 20% off the list price.
LLM API Model Families
Each family page lists every available version with its context window, token pricing and a playground to test prompts before you integrate.
DeepSeek API
DeepSeek V4 API access to DeepSeek's open-source model family — V4 Pro, V4 Flash and V4.1 Flash chat models plus the R1 reasoning model, delivering frontier-level performance at a fraction of the cost.
ReasoningOpen SourceCost-EffectiveView models & pricing →Qwen API
The Qwen family covers general and multimodal workloads. This page highlights qwen3.8-max, with up to 983,616 input tokens, 131,072 output tokens, required reasoning, caching, and built-in tools through the Responses API.
Long ContextRequired ReasoningBuilt-in ToolsView models & pricing →Doubao API
ByteDance's Doubao model family — offering strong Chinese-English capabilities, vision understanding, and advanced thinking modes for complex tasks.
Chinese NLPVisionThinkingView models & pricing →GLM API
Zhipu AI's GLM model family — Chinese-English bilingual models with rapid iteration from GLM-4.6 to GLM-5.3, delivering strong reasoning and generation quality.
BilingualReasoningChatView models & pricing →Kimi API
Kimi K3 API access to Moonshot AI's Kimi model family — K3 and K2.7 Code with strong reasoning, coding capabilities, and extended thinking modes.
ReasoningCodingThinkingView models & pricing →MiniMax API
MiniMax's language model family — from M2.1 to M3, offering competitive performance for general chat, content creation, and bilingual tasks.
ChatContent CreationBilingualView models & pricing →
How to Choose an LLM API
Complex reasoning and coding
Qwen 3.8 Max, DeepSeek V4.1 and Kimi K3 handle multi-step reasoning, agents and large codebases.
Fast and low-cost chat
DeepSeek V4.1 Flash, GLM-5.3 Flash and Doubao models keep latency and token cost low for high-volume chat and extraction.
Long context
Several models support very long context windows for document analysis and retrieval; check each model's context limit on its family page.
Drop-in OpenAI compatibility
Use the OpenAI SDK you already have: point the base URL to 光谱汇深 and change the model name to switch providers.
LLM API FAQ
Is the 光谱汇深 LLM API OpenAI-compatible?
Yes. Chat models use the OpenAI-compatible Chat Completions format, so existing OpenAI SDK code works after changing the base URL, API key and model name.
How is LLM API pricing calculated?
LLM models are billed per million input and output tokens, and some models also price cached input. Each family page shows the token pricing per model, and every model is 20% off the list price, with membership tiers saving up to 28%.
Which LLM APIs are available?
光谱汇深 offers chat models from DeepSeek, Alibaba Qwen, ByteDance Doubao, Zhipu GLM, Moonshot Kimi and MiniMax, with new versions added as they launch.
Can I use one API key for multiple LLM providers?
Yes. A single 光谱汇深 key works for every chat, image and video model, with one balance and one invoice.
Do you support streaming and function calling?
Yes. Streaming responses, function or tool calling and vision input are supported on models whose providers offer those features.