AI Text APIs
AI Text APIs on HiAPI
Section titled “AI Text APIs on HiAPI”HiAPI exposes text models through OpenAI-compatible endpoints while keeping the same account API key used for media generation.
| Model | Endpoint | Best for | Key capabilities |
|---|---|---|---|
| Claude Opus 5.5 | POST /v1/chat/completions | Long-running agentic coding and knowledge work | Always-on thinking; low/medium/high/xhigh/max effort; medium default |
| Claude Haiku 5.5 | POST /v1/chat/completions | Fast text and coding workflows | Native reasoning_effort; low/medium/high/xhigh; medium default |
| Grok 4.6 | POST /v1/chat/completions | Reasoning, streaming, JSON objects, and tools | low/medium/high/xhigh; high default |
| Grok 4.7 | POST /v1/chat/completions | Reasoning, streaming, JSON objects, and tools | low/medium/high/xhigh; high default |
| GPT-6 Astra | POST /v1/responses · streaming only | Reasoning, code review, function workflows | low/medium/high/xhigh/max reasoning, strict JSON Schema, function calls |
| GPT-6.1 Sol | POST /v1/responses · streaming only | Reasoning, coding, and analysis | low/medium/high/xhigh/max reasoning |
| GPT-6.1 Terra | POST /v1/responses + Chat Completions | Visual reasoning, computer use, and office tasks | Image input, none/low/medium/high/xhigh/max reasoning, function calls, strict JSON schema |
| GPT-6 Luna | POST /v1/responses + Chat Completions | Summaries, extraction, and high-volume quick Q&A | Image input, none/low/medium/high/xhigh/max reasoning, function calls |
| Claude Sonnet 4.6 | POST /v1/chat/completions | Text conversations, coding, and tool workflows | Streaming and OpenAI-compatible tools |
| Kimi K3.1 Pro | POST /v1/chat/completions | Coding, agent tool workflows, and long documents | 1M context, low/high/max reasoning, streaming, JSON, function tools |
| Kimi K3.1 Pro Highspeed | POST /v1/chat/completions | Latency-sensitive coding, agent tool workflows, and long documents | Faster K3.1 Pro variant, 1M context, low/high/max reasoning, streaming, JSON, function tools |
| Kimi K3 | POST /v1/chat/completions | Long-context reasoning, coding, and tool workflows | 1M context, low/high/max reasoning, streaming, JSON, tools |
| DeepSeek V4.1 Flash | POST /v1/responses + Chat | Text and image input, reasoning | none/low/high/max, SSE |
| DeepSeek V4 Pro | POST /v1/responses + Chat compatible | Complex reasoning, coding Agents, long context | 1M context, 384K max output, none/low/high/max reasoning, JSON, tools |
| DeepSeek V4 Flash Vision Exp | POST /v1/responses + Chat compatible | Image understanding, visual questions, function tools | 1,048,576 context, 262,144 output limit, none/low/medium/high reasoning, JSON |
| DeepSeek V4 Flash | POST /v1/chat/completions | High-throughput generation, reasoning, coding, Agent workflows | 1M context, streaming, high/max reasoning, JSON mode, tool calls |
| GLM-5.3 | POST /v1/chat/completions | Text generation through an OpenAI-compatible endpoint | See the model page for the current request contract |
Choose the protocol first
Section titled “Choose the protocol first”- New Agent and Codex workflows:
POST /v1/responseswhen listed on the model page. - Existing chat clients: compatible
POST /v1/chat/completions. - Image, video, and audio generation: asynchronous
POST /v1/tasks.
Use live pricing for current input, output, and cached-input Token rates.
Get a HiAPI API key or start with the Responses guide. Existing clients can continue with Chat Completions.