OpenPAI fully integrates every available OpenAI model and keeps the model IDs consistent with the official ones.
For live prices, refer to Console → Models.
Chat models
| Model ID | Category | Context | Input price (per 1M tokens) | Output price (per 1M tokens) | Key features |
|---|
gpt-5.5 | Latest flagship | 1M | See console | See console | A new class of intelligence for coding and professional work |
gpt-5.4 | Flagship (more affordable) | 1M | See console | See console | High-quality code and general chat, better value |
gpt-5.4-mini | Economical | 400K | See console | See console | Strongest mini for code / computer use / subagents |
gpt-5.4-nano | Fast & economical | — | — | — | Low-latency, low-cost, high-volume workloads |
gpt-4o | Older flagship multimodal | 128K | See console | See console | Vision, tools, JSON, streaming |
gpt-4o-mini | Older economical | 128K | See console | See console | Still available for legacy projects |
gpt-3.5-turbo | Older economical | 16K | — | — | Recommended only for legacy compatibility |
OpenAI’s chat capabilities are increasingly centered on the
GPT-5.x series, and OpenAI recommends the
Responses API for new projects. The GPT-4o series remains callable.
Reasoning models (GPT-5.x / o series)
The GPT-5.x series has built-in reasoning, controlled via reasoning_effort; the o series (o3, o4-mini, etc.) is still available.
| Model ID / parameter | Reasoning effort | Use case |
|---|
gpt-5.5 + reasoning_effort: low | Low | Simple reasoning, fast responses |
gpt-5.5 + reasoning_effort: medium | Medium | Balance of quality and speed |
gpt-5.5 + reasoning_effort: high | High | Complex logic, math, code |
o4-mini | Medium-high | Economical reasoning |
o3 | High | Strong reasoning (previous-gen flagship reasoning) |
Reasoning models are billed including invisible reasoning_tokens — they aren’t shown in content but count toward usage. Leave headroom when estimating max_completion_tokens.
resp = client.chat.completions.create(
model="gpt-5.5",
messages=[{"role": "user", "content": "Prove that √2 is irrational"}],
reasoning_effort="high",
max_completion_tokens=4096,
)
Image generation
| Model ID | Type | Notes |
|---|
gpt-image-2 | Image generation | Latest generation, highest quality, supports multi-image editing |
gpt-image-1 | Image generation | Previous generation, high quality, supports multi-image editing |
dall-e-3 | Image generation | Classic DALL·E 3, sizes like 1024 / 1792 |
dall-e-2 | Image generation | Older, cheaper |
img = client.images.generate(
model="gpt-image-2",
prompt="A black-and-white pencil sketch of a Shiba Inu wearing glasses",
size="1024x1024",
n=1,
)
print(img.data[0].url)
Audio
| Model ID | Use case |
|---|
whisper-1 | Speech to text (ASR) |
gpt-4o-transcribe | High-quality transcription |
gpt-4o-mini-transcribe | Economical transcription |
tts-1 | Text to speech (standard) |
tts-1-hd | Text to speech (hi-fi) |
gpt-4o-mini-tts | Next-gen TTS, more natural voice |
curl https://openp.ai/v1/audio/speech \
-H "Authorization: Bearer $OPENPAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"tts-1","input":"Hello from OpenPAI","voice":"alloy"}' \
--output speech.mp3
Realtime (voice conversation)
| Model ID | Use case |
|---|
gpt-realtime-2 | Latest realtime voice (with reasoning) |
gpt-realtime-1.5 | High-quality audio in / audio out |
gpt-realtime-translate | Realtime speech-to-speech translation |
gpt-realtime-whisper | Streaming speech-to-text |
gpt-4o-realtime-preview | Previous-gen realtime multimodal conversation |
wss://openp.ai/v1/realtime?model=gpt-realtime-2, over WebSocket.
See the Realtime API docs.
Embeddings
| Model ID | Dimensions | Main scenario |
|---|
text-embedding-3-small | 1536 (reducible) | General, default recommendation |
text-embedding-3-large | 3072 (reducible) | High-quality retrieval |
text-embedding-ada-002 | 1536 | Legacy compatibility |
Moderations
omni-moderation-latest — multimodal content-safety classification, free (some accounts must enable it).
Responses API
OpenAI’s newer /v1/responses endpoint unifies conversation state management, suited to agent scenarios.
See the Responses API.
Detailed fields
