The DeepSeek V4 Flash is now directly compatible with Codex!
You can keep using Codex's interface, tools, and workflows, but swap out the "brain" for DeepSeek.
For every million tokens in the V4 Flash:
- Input: $0.14
- Output: $0.28
Let's say you consume 5 million input tokens and 500,000 output tokens in a day. Assuming all cache misses, I only need to pay about $1 .
If a project goes on for three days, using DeepSeek might cost less than $3, whereas other models could set you back a couple of hundred.
Before, I would worry about it repeatedly reading files, correcting errors, and wasting tokens. Now, all I want to say is:
Go ahead boy, burn my money however you like.
By the way, the V4 Pro is expected to support Codex by early August as well.