GPT-5.4 mini and nano
GPT-5.4 mini and nano are OpenAI's latest compact language models, released March 2026.
PaidVisit site ↗
Our verdict
Who it's for
Compact language models for coding, reasoning, and data tasks
Who should look elsewhere
August 2025 knowledge cutoff means no awareness of recent events
Pros
- +GPT-5.4 mini approaches full GPT-5.4 performance at a fraction of the cost
- +Nano model at $0.20/1M input tokens is extremely cost-effective for high-volume tasks
- +400K context window handles large documents and codebases
- +2x faster than previous generation mini models
- +Available across ChatGPT (free tier), Codex, and API
Cons
- −August 2025 knowledge cutoff means no awareness of recent events
- −Nano lacks some advanced capabilities like computer use
- −Text-only output (no native image or audio generation)
- −Mini at $4.50/1M output tokens can get expensive for verbose tasks
- −Smaller models may still hallucinate more than full GPT-5.4
Key features
- ✓400,000-token context window on both models
- ✓128,000 maximum output tokens
- ✓Text and image input support
- ✓Function calling and structured outputs
- ✓Web search, file search, and code interpretation tools
- ✓Streaming and batch processing support
- ✓Computer use capability (mini)
- ✓Reasoning token support for chain-of-thought
- ✓Realtime API compatibility
- ✓90% discount on cached input tokens
Pricing
No public pricing plans recorded.