Vendor · OpenAI · 15 models
OpenAI API on Crazyrouter
OpenAI's API is card-only and unavailable in several regions. Crazyrouter serves the same models through `/v1/chat/completions` and the native `/v1/responses` endpoint (reasoning items, encrypted reasoning and tool calls pass through unchanged), bills per token at 35% below list for the frontier models, and works with Codex CLI, Cursor and the OpenAI SDKs by changing `base_url`.
Cheapest OpenAI model here: GPT-4o mini at $0.0975 / $0.39 per 1M tokens. Up to 35% below list.

Chat and reasoning models
OpenAI
−35%
below list
Input / 1M$1.25$0.813
Output / 1M$10$6.50
Context 272K tokensCached $0.085Also on Azure
GPT-5
gpt-5Crazyrouter price: $0.813 / $6.50 per 1M tokens · list $1.25 / $10.00
OpenAI compatibleReasoningCoding
OpenAI
−35%
below list
Input / 1M$5.00$3.25
Output / 1M$30$20
Context 1M tokensCached $0.325Also on Azure
GPT-5.5
gpt-5.5Crazyrouter price: $3.25 / $19.50 per 1M tokens · list $5.00 / $30.00
OpenAI compatibleReasoningCoding
OpenAI
−35%
below list
Input / 1M$2.50$1.63
Output / 1M$15$9.75
Context 1M tokensCached $0.163Also on Azure
GPT-5.4
gpt-5.4Crazyrouter price: $1.63 / $9.75 per 1M tokens · list $2.50 / $15.00
OpenAI compatibleReasoningCoding
OpenAI
−35%
below list
Input / 1M$0.750$0.488
Output / 1M$4.50$2.93
Context 272K tokensCached $0.049Also on Azure
GPT-5.4 mini
gpt-5.4-miniCrazyrouter price: $0.488 / $2.93 per 1M tokens · list $0.75 / $4.50
OpenAI compatibleFastLow cost
OpenAI
−35%
below list
Input / 1M$4.00$3.25
Output / 1M$20$20
Context 922K tokensCached $0.325Also on Azure
GPT-5.6 Sol
gpt-5.6-solCrazyrouter price: $3.25 / $19.50 per 1M tokens · list $4.00 / $20.00
OpenAI compatibleReasoningCoding
OpenAI
−35%
below list
Input / 1M$10$3.25
Output / 1M$50$16
Context 922K tokensCached $0.325Also on Azure
GPT-6 Astra
gpt-6-astraCrazyrouter price: $3.25 / $16.25 per 1M tokens · list $10.00 / $50.00
OpenAI compatibleReasoningCoding
OpenAI
−35%
below list
Input / 1M$1.75$1.14
Output / 1M$14$9.10
Context 272K tokensCached $0.114Also on Azure
GPT-5.3 Codex
gpt-5.3-codexCrazyrouter price: $1.14 / $9.10 per 1M tokens · list $1.75 / $14.00
OpenAI compatibleCoding agentsCodex CLI
OpenAI
−35%
below list
Input / 1M$1.25$0.813
Output / 1M$10$6.50
Context 272K tokensCached $0.085Also on Azure
GPT-5 Codex
gpt-5-codexCrazyrouter price: $0.813 / $6.50 per 1M tokens · list $1.25 / $10.00
OpenAI compatibleCoding agentsCodex CLI
OpenAI
−35%
below list
Input / 1M$2.00$1.30
Output / 1M$8.00$5.20
Context 1M tokensCached $0.325Also on Azure
GPT-4.1
gpt-4.1Crazyrouter price: $1.30 / $5.20 per 1M tokens · list $2.00 / $8.00
OpenAI compatibleReasoningCoding
OpenAI
−35%
below list
Input / 1M$0.400$0.260
Output / 1M$1.60$1.04
Context 1M tokensCached $0.065Also on Azure
GPT-4.1 mini
gpt-4.1-miniCrazyrouter price: $0.26 / $1.04 per 1M tokens · list $0.40 / $1.60
OpenAI compatibleFastLow cost
OpenAI
−35%
below list
Input / 1M$2.50$1.63
Output / 1M$10$6.50
Context 128K tokensCached $0.813Also on Azure
GPT-4o
gpt-4oCrazyrouter price: $1.63 / $6.50 per 1M tokens · list $2.50 / $10.00
OpenAI compatibleReasoningCoding
OpenAI
−35%
below list
Input / 1M$0.150$0.098
Output / 1M$0.600$0.390
Context 128K tokensCached $0.052Also on Azure
GPT-4o mini
gpt-4o-miniCrazyrouter price: $0.0975 / $0.39 per 1M tokens · list $0.15 / $0.60
OpenAI compatibleFastLow cost
OpenAI
Input / 1M$2.00$2.00
Output / 1M$8.00$8.00
Context 200K tokensCached $0.500Also on Azure
OpenAI o3
o3Crazyrouter price: $2.00 / $8.00 per 1M tokens · list $2.00 / $8.00
OpenAI compatibleReasoningMath
Image models


GPT Image 1.5
gpt-image-1.5Crazyrouter price: $0.107 / image
Image generationText to imageImage editing
Why route OpenAI through Crazyrouter
- Native Responses API with encrypted reasoning, so Codex CLI and agent frameworks behave as on OpenAI.
- GPT-6 Astra, GPT-5.6 Sol and GPT-5.5 at 35% below list; GPT-5 and GPT-4.1 for volume work.
- GPT Image 2 and GPT Image 1.5 behind the same key — see the image samples on each model page.
- Multiple upstream channels with priority routing; failed requests are retried on another channel before you see an error.
FAQ
- Which endpoint should I use?
- `/v1/chat/completions` for anything OpenAI-compatible; `/v1/responses` when you need reasoning items, encrypted reasoning or the Codex CLI. Both accept the same key.
- Is GPT-6 Astra available?
- Yes, at $3.25 / $16.25 per 1M tokens against OpenAI's $10 / $50. It is the most-routed model on the platform.
- Does Codex CLI work?
- Yes — set the provider base URL to `https://api.crazyrouter.com/v1` in `config.toml`; the one-click script on the Codex integration page does it for you.
- Are image models billed per token?
- GPT Image models are billed per image by size and quality, shown on each model page; text models are per token.