OpenAI GPT
OpenAI's GPT and o-series reasoning models, covering coding, reasoning, agents and everyday text tasks. GPT Image is for image generation and editing.
Choosing a group
| Group | Price |
|---|---|
az_gpt | Pay 30% of official price |
codex_gpt | Pay 4.45% of official price |
gpt | Pay 10% of official price |
gpt_image | Pay 10% of official price |
gpt,codex_gpt: the latest main GPT models.az_gpt: the most complete set of models, including older ones such as GPT-4o and GPT-4.1, and thetext-embedding-3embedding models.gpt_image: GPT Image models. See Image generation and Image edits for usage.
Recommended API
- All text models work with Chat Completions.
- Models marked Responses also work with Responses, which Codex CLI requires.
bash
curl https://api.codingrouter.ai/v1/chat/completions \
-H "Authorization: Bearer $CODINGROUTER_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-6.1-sol",
"reasoning_effort": "medium",
"messages": [{"role": "user", "content": "Check this SQL for injection risks: ..."}]
}'Usage tips
- Reasoning models use
reasoning_effort(Chat Completions) orreasoning.effort(Responses) to control thinking depth, with valueslow,medium,high. The thinking process counts as output tokens.
Representative models
| Model | Group | Good for |
|---|---|---|
gpt-6.1-sol | gpt, codex_gpt, az_gpt | Complex coding and agents, supports Responses, recommended for Codex CLI |
gpt-6-astra | gpt, codex_gpt, az_gpt | Hard reasoning |
gpt-5.5 | gpt, codex_gpt, az_gpt | Coding, reasoning, multimodal understanding |
gpt-5.4-mini | az_gpt | Low-cost everyday tasks |
text-embedding-3-small / -large | az_gpt | Embeddings |
gpt-image-2 / gpt-image-2.5 | gpt_image | Image generation and editing |
For the full list and prices see the model square.
