GPT-5.4 launches on ChatGPT with gains in efficiency and accuracy

OpenAI released GPT‑5.4 this week, calling it the "most capable and efficient frontier model for professional work." The launch follows GPT‑5.3 Instant and includes GPT‑5.4 Thinking and GPT‑5.4 Pro for users seeking peak performance. OpenAI said GPT‑5.4 is its most token-efficient reasoning model yet, using significantly fewer tokens than GPT‑5.2 to solve problems, translating to reduced token usage and faster speeds.

The company also reported accuracy gains: the model is 33 percent less likely to make false claims and 18 percent less likely to contain any errors compared with GPT‑5.2. On real-world tasks, GPT‑5.4 set a new state of the art on GDPval, a benchmark for knowledge work across 44 occupations, matching or exceeding industry professionals in 83.0 percent of comparisons versus 70.9 percent for GPT‑5.2.

GPT‑5.4 Thinking can offer an upfront plan of its reasoning so users can adjust course mid-response and arrive at outputs that more closely match their needs.

gpt-5.4, gpt-5.3 instant, gpt-5.4 thinking, gpt-5.4 pro, token efficiency, reasoning model, accuracy gains, false claims, gdpval benchmark, knowledge work