Model comparison
Claude Haiku 4.5 vs GPT-5.3 Codex
List rate, corouter rate, context, and which ID to send for claude-haiku-4-5 versus gpt-5.3-codex. A sample of 1 million input tokens plus 200 thousand output tokens is about $0.400 on corouter (input $0.200, output $0.200). A sample of 1 million input tokens plus 200 thousand output tokens is about $0.910 on corouter (input $0.350, output $0.560).
| Claude Haiku 4.5 | GPT-5.3 Codex | |
|---|---|---|
| ID | claude-haiku-4-5 | gpt-5.3-codex |
| Vendor | Anthropic | OpenAI |
| List input | $1.00 | $1.75 |
| corouter input | $0.200 | $0.350 |
| List output | $5.00 | $14.00 |
| corouter output | $1.00 | $2.80 |
| Cache read | $0.020 | $0.035 |
| Context | 200K | 272K |
Price gap
On output, Claude Haiku 4.5 is lower: $1.00 on corouter versus $2.80 for GPT-5.3 Codex, about 2.80×. Input is $0.200 versus $0.350.
Window and modalities
GPT-5.3 Codex lists a 272K window and Claude Haiku 4.5 lists 200K. A larger window does not change the per-token rate. Prompts past the window need a split or a different ID. Claude Haiku 4.5 is listed with these modalities: Chat, Vision. GPT-5.3 Codex is listed with these modalities: Chat, Reasoning, Code, Vision.
When to send claude-haiku-4-5
Claude Haiku 4.5 is the short-window, low list-rate Claude row: 200K, chat and vision, no Code tag. Send claude-haiku-4-5 for volume that fits that window.
This ID uses the familiar messages JSON shape. The other side is gpt-5.3-codex from OpenAI. corouter is not affiliated with OpenAI or Anthropic.
When to send gpt-5.3-codex
GPT-5.3 Codex is a coding ID with a 272K window. Its list rate matches GPT-5.2 Codex; the difference you bill is the ID itself, gpt-5.3-codex.
This ID uses the familiar chat-completions JSON shape. The other side is claude-haiku-4-5 from Anthropic. corouter is not affiliated with OpenAI or Anthropic.
The same volume, two bills
A sample of 1 million input tokens plus 200 thousand output tokens is about $0.400 on corouter (input $0.200, output $0.200).
A sample of 1 million input tokens plus 200 thousand output tokens is about $0.910 on corouter (input $0.350, output $0.560).
Questions
- Which output rate is lower, Claude Haiku 4.5 or GPT-5.3 Codex?
- On output, Claude Haiku 4.5 is lower: $1.00 on corouter versus $2.80 for GPT-5.3 Codex, about 2.80×. Input is $0.200 versus $0.350.
- Is corouter affiliated with the model vendors?
- No. corouter is not affiliated with OpenAI or Anthropic. Vendor names identify the model and the published list rate.
- How do you start billing?
- Create a console account and spend prepaid credits. The minimum top-up is $5. The API base is https://api.corouter.co. Put the ID you chose in the model field.