GPT-5.4 mini
OpenAI’s strongest mini-class model for coding, computer use, and subagent workloads where price and latency matter.
Capability profile
- Input
- $0.75 / 1M tokens
- Output
- $4.50 / 1M tokens
- Cached input
- $0.075 / 1M tokens
- Context
- 400K tokens
- Free tier
- No / unknown
- API
- Available
- Subscription
- OpenAI API
- Open-source status
- Closed
- High-volume agent backends
- coding subagents
- and practical production automation.
- Subagents
- coding assistants
- tool-using automations
- and scalable internal products.
- Truly frontier reasoning tasks where the top full-size models pull away.
- The hardest research-grade reasoning or premium polished writing as a primary use.
This is the kind of model that actually changes unit economics. It makes a lot of agent patterns viable where flagship pricing still feels heavy.
OpenAI’s strongest mini-class model for coding, computer use, and subagent workloads where price and latency matter.
Post Reboot take
This is the kind of model that actually changes unit economics. It makes a lot of agent patterns viable where flagship pricing still feels heavy.
Strengths
Best for: High-volume agent backends, coding subagents, and practical production automation.
Weakest at: Truly frontier reasoning tasks where the top full-size models pull away.
Best workflows: Subagents, coding assistants, tool-using automations, and scalable internal products.
Example use cases: Bulk code transformations, workflow orchestration, QA triage, and medium-complexity app assistance.