GPT-5.4 mini

OpenAIMULTIMODAL Proprietary
Provider
OpenAI
Modality
MULTIMODAL
Parameters
Context window
1,050,000 tokens
Weights
Proprietary
Released
17 Mar 2026
Compare GPT-5.4 mini with others

GPT-5.4 mini is the cost-efficient workhorse — cheap, fast and good enough for the bulk of everyday traffic, the free/fallback tier in ChatGPT.

Best for

  • High-volume production traffic where cost-per-call dominates
  • Chat, summarisation, classification and routing
  • Latency-sensitive features that don't need flagship reasoning

How it compares — and the India angle

  • Best price/performance in the family — route the majority of an India-focused app's requests here and escalate only when quality demands it.
  • Its ~54% SWE-bench means it's not a serious autonomous coder; don't use it for hard multi-step bug-fixing.
  • Cheaper still: GPT-5.4 nano for pure routing/classification.

Benchmarks

Representative public scores (approximate, higher is better) for relative comparison. Check the provider for the latest official results.

SWE-bench Verified54

How to access

API ~$0.75 / 1M input, ~$4.50 / 1M output (sources vary). The free-tier fallback in ChatGPT.

Access GPT-5.4 mini
Share: