| Model ID | Modalities | Notes |
|---|---|---|
gpt-4o | textvision | Flagship versatile model. Strong text + vision. (See docs) |
gpt-4.1 | text | High accuracy, long context. (See docs) |
gpt-4o-realtime-preview | realtimeaudiotext | Realtime via WebRTC / WebSocket. (See docs) |
gpt-4o-mini | textvision | Smaller, fast, low-cost omni. (See docs) |
gpt-4o-mini-realtime-preview | realtimeaudio | Realtime small omni. (See docs) |
| Model ID | Modalities | Notes |
|---|---|---|
o3 | textvisionreasoning | High-power reasoning; math/science/coding. (OpenAI docs) |
o3-pro | textreasoning | “Think harder” variant; Responses API. (OpenAI docs) |
o3-mini | textreasoning | Small reasoning model; fast + lower cost. (OpenAI docs) |
o3-deep-research | textweb | Automated multi-source research. (Deep‑research docs) |
o4-mini | textreasoning | Fast reasoning for STEM/coding. (OpenAI docs) |
o4-mini-deep-research | textweb | Cheaper deep-research agent. (OpenAI docs) |
| Model ID | Modalities | Notes |
|---|---|---|
gpt-4o-audio-preview | audiotext | Audio in/out (preview). (OpenAI docs) |
gpt-4o-mini-audio-preview | audiotext | Smaller audio model (preview). (OpenAI docs) |
gpt-4o-mini-tts | tts | Text‑to‑speech. (OpenAI docs) |
gpt-4o-mini-transcribe | stt | Speech‑to‑text. (OpenAI docs) |
gpt-image-1 | image-gen | Image generation. (OpenAI docs) |
Docs references: GPT‑4.1, GPT‑4o, 4o‑mini, Realtime, Audio/Transcribe, o3/o4 families. Always consult the official docs for real‑time availability and previews.
Sources: OpenAI model pages for GPT‑4.1, GPT‑4o, 4o‑mini, 4o Realtime, 4o‑mini TTS, 4o‑mini Transcribe, o3, o3‑mini, o3‑pro, o4‑mini, and deep‑research.
Groq
| Model ID | Size / Window | Notes |
|---|---|---|
openai/gpt-oss-20b | ~20B · 131k ctx | OpenAI open‑weight model hosted on Groq. (Groq docs) |
openai/gpt-oss-120b | ~120B · 131k ctx | OpenAI open‑weight flagship hosted on Groq. (Groq docs) |
llama-3.3-70b-versatile | 70B · 131k ctx | Latest Meta Llama family variant on Groq. (Groq docs) |
llama-3.1-8b-instant | 8B · 131k ctx | Fast, small Llama 3.1. (Groq docs) |
llama3-70b-8192 | 70B · 8k ctx | Stable Llama 3 (OpenAI‑compatible id). (Groq docs) |
llama3-8b-8192 | 8B · 8k ctx | Small Llama 3 (OpenAI‑compatible id). (Groq docs) |
mixtral-8x7b-32768 | MoE · 32k ctx | Mistral Mixture‑of‑Experts. (Groq docs) |
gemma2-9b-it | 9B | Google Gemma 2 instruct. (Groq docs) |
| Model ID | Type | Notes |
|---|---|---|
whisper-large-v3 | STT | OpenAI Whisper hosted by Groq. (Groq docs) |
whisper-large-v3-turbo | STT | Faster Whisper. (Groq docs) |
Groq’s page includes live model IDs and a JSON endpoint to enumerate all active models. Some entries (e.g., preview systems) can change frequently.
Source: Groq docs: Supported Models (lists GPT‑OSS 20B/120B, Llama 3.3‑70B, 3.1‑8B, Whisper Large V3, etc.).
Google Gemini
| Model ID | Modalities | Notes |
|---|---|---|
gemini-1.5-pro / gemini-1.5-pro-002 | textvisionaudio | Multimodal, long context (up to ~1M tokens). (Google docs) |
gemini-1.5-flash / gemini-1.5-flash-002 | textvisionaudio | Fast, low‑latency multimodal. (Google docs) |
gemini-1.5-flash-8b | textvisionaudio | Small, efficient multimodal. (Google docs) |
Some model versions can be restricted or temporarily unavailable in new projects (see Google’s model lifecycle notes).
Sources: Gemini models page; Gemini 1.5 Flash and Flash‑8B specs; changelog entries noting new 1.5 Pro/Flash versions.
Other ecosystems (FYI)
Not wired into the UI by default here, but often requested:
- Anthropic Claude family (e.g., Claude 3.5/4.* lines)—see Anthropic model overview.
- xAI Grok 4 and newer—see xAI API docs (OpenAI‑style endpoints).
Docs: Anthropic Models, xAI API.