gpt
Call GPT generation and OpenAI text embedding models through RunAPI using OpenAI-compatible clients. Use for chat, Responses, embeddings, streaming, tools, vision, or an existing c…
RunAPI
@runapi-ai
What This Skill Does
Provides a unified API endpoint to call GPT-5 series chat, reasoning, and Codex models, plus OpenAI text embeddings, through any OpenAI-compatible SDK by simply changing the base URL and API key. Supports Chat Completions, Responses API, and Embeddings endpoints with no client code changes beyond configuration.
Replaces managing separate API keys and endpoints for each OpenAI model family by routing all GPT, reasoning, Codex, and embedding calls through a single RunAPI base URL.
When to Use It
- Call GPT-5 chat models from Python or TypeScript using the official OpenAI SDK
- Generate text embeddings for semantic search or vector databases
- Use reasoning models with configurable effort levels (low/medium/high)
- Run Codex coding models for code generation or completion tasks
- Stream chat completions from GPT models in real-time applications
- Switch an existing OpenAI SDK integration to RunAPI by changing only the base URL and API key
Install
$ openclaw skills install @runapi-ai/runapi-gptGPT on RunAPI
Use OpenAI-compatible clients at https://runapi.ai/v1 as the primary protocol.
Primary protocol recipe
Authenticate
Set OPENAI_API_KEY to a RunAPI API key and OPENAI_BASE_URL to https://runapi.ai/v1.
Send request
from openai import OpenAI
client = OpenAI(api_key="YOUR_RUNAPI_TOKEN", base_url="https://runapi.ai/v1")
response = client.responses.create(model="gpt-5.5", input="Summarize this review.")
print(response.output_text)
print(response.usage)
Use client.chat.completions.create for gpt-5.2, which does not accept
Responses. Use Responses for gpt-5.*-pro. Use client.embeddings.create only
with the three embedding model IDs below. For streaming Responses, set
stream=True and consume through one usage-bearing response.completed
followed by [DONE].
For gpt-5.6-luna, gpt-5.6-sol, and gpt-5.6-terra, start with text input,
sync or SSE transport, parameterized custom functions, and automatic prompt
caching. gpt-5.3-codex-spark has the same verified subset with parameterless
custom functions. Add reasoning controls, multimodal input, hosted tools, or
continuation state only when the current RunAPI contract verifies that shape.
Verify result
Responses require final output, response.completed, and usage. Chat requires
final assistant content, finish_reason, and usage. Embeddings require one
numeric vector per input and response usage.
Stop boundaries
Correct a rejected shape once using the current error and endpoint contract.
Retry transport once only before any response or Usage and when replay is safe.
Record a terminal error and stop without changing model, endpoint, or protocol.
Keep embedding models on /v1/embeddings.
Compatibility protocols
Load compatibility protocols only when an existing client requires Anthropic Messages or Gemini contents.
Supported models
| Model ID | Use when |
|---|---|
gpt-5.5 | Latest general model |
gpt-5.5-pro | Reasoning-heavy |
gpt-5.4 | Production default |
gpt-5.4-mini | Cost-optimized |
gpt-5.4-nano | Smallest, fastest |
gpt-5.4-pro | Reasoning |
gpt-5.3-codex | Code generation |
gpt-5.3-codex-spark | Faster Codex variant; text and parameterless functions in the consistently available Responses subset |
gpt-5.2 | Cost-effective |
gpt-5.6-luna | Text and parameterized functions in the consistently available Responses subset |
gpt-5.6-sol | Text and parameterized functions in the consistently available Responses subset |
gpt-5.6-terra | Text and parameterized functions in the consistently available Responses subset |
text-embedding-3-large | High-capacity vectors |
text-embedding-3-small | Efficient vectors |
text-embedding-ada-002 | Legacy-compatible vectors |
References
Top skills in this category
Humanizer
@biostartechnologyRemove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, negative parallelisms, and excessive conjunctive phrases.
Elite Longterm Memory
@nextfrontierbuildsUltimate AI agent memory system for Cursor, Claude, ChatGPT & Copilot. WAL protocol + vector search + git-notes + cloud backup. Never lose context again. Vibe-coding ready.
Model Usage
@steipeteUse CodexBar CLI local cost usage to summarize per-model usage for Codex or Claude, including the current (most recent) model or a full model breakdown. Trigger when asked for model-level usage/cost data from codexbar, or when you need a scriptable per-model summary from codexbar cost JSON.
Screenshot
@ivangdavilaCapture, inspect, and compare screenshots of screens, windows, regions, web pages, simulators, and CI runs with the right tool, wait strategy, viewport, and...
Interview Simulator
@wscatsSimulates mock interviews for any role and experience level with tailored technical, behavioral, and case questions plus detailed feedback and scoring.