OpenAI says /fast mode delivers up to 1.5× faster performance across supported models, including GPT-5.4, describing it as the same model and intelligence “just faster.” And it describes releasing an experimental Codex…
On MMMU-Pro, GPT-5.4 reaches 81.2% success without tool use, compared with 79.5% for GPT-5.2, and OpenAI says it achieves that result using a fraction of the “thinking tokens.” On OmniDocBench, GPT-5.4’s average error i…
ChatGPT Free users will also get a taste of GPT-5.4, but only when their queries are auto-routed to the model, according to an OpenAI spokesperson.
Pricing and availabilityIn the API, OpenAI says GPT-5.4 Thinking is available as gpt-5.4 and GPT-5.4 Pro as gpt-5.4-pro.
Ключевые утверждения в источнике B
Daniel Swiecki of Walleye Capital said GPT-5.4 “improved accuracy by 30 percentage points” on internal finance and Excel evaluations, a VentureBeat noted.
Agentic Performance: The model achieves a 75.0% success rate on OSWorld-Verified, surpassing the reported human performance baseline of 72.4% and up from 47.3% for GPT-5.2.
the model achieves a 75.0% success rate on OSWorld-Verified, up from 47.3% for GPT-5.2 and above the 72.4% reported human performance baseline.
On web navigation benchmarks, OpenAI said the model reaches 67.3% on the WebArena-Verified benchmark, with 92.8% on Online-Mind2Web using screenshot-based observations.
Текстовые доказательства
Доказательства из источника A
ключевое утверждение
ChatGPT Free users will also get a taste of GPT-5.4, but only when their queries are auto-routed to the model, according to an OpenAI spokesperson.
Ключевое утверждение, на котором строится подача материала.
ключевое утверждение
OpenAI says /fast mode delivers up to 1.5× faster performance across supported models, including GPT-5.4, describing it as the same model and intelligence “just faster.” And it describes re…
Ключевое утверждение, на котором строится подача материала.
причинно-следственная связка
OpenAI’s emphasis on token efficiency, tool search, native computer use, and reduced user-flagged factual errors all point in the same direction: making agentic systems more viable in produ…
Причинно-следственное утверждение, влияющее на объяснение событий.
возможное упущение контекста
Daniel Swiecki of Walleye Capital said GPT-5.4 “improved accuracy by 30 percentage points” on internal finance and Excel evaluations, a VentureBeat noted.
Возможное различие контекста: источник A слабее раскрывает блок о экономическим и ресурсным факторам, чем источник B.
Доказательства из источника B
ключевое утверждение
Daniel Swiecki of Walleye Capital said GPT-5.4 “improved accuracy by 30 percentage points” on internal finance and Excel evaluations, a VentureBeat noted.
Ключевое утверждение, на котором строится подача материала.
ключевое утверждение
Agentic Performance: The model achieves a 75.0% success rate on OSWorld-Verified, surpassing the reported human performance baseline of 72.4% and up from 47.3% for GPT-5.2.
Ключевое утверждение, на котором строится подача материала.
причинно-следственная связка
Tool yields are a better proxy of latency than tool calls because they reflect the benefits of parallelization.
Причинно-следственное утверждение, влияющее на объяснение событий.
выборочный акцент
Available in two variants, GPT-5.4 Thinking and GPT-5.4 Pro, the model unifies reasoning, coding, and agentic workflows into a single release arriving just two days after GPT-5.3 Instant.
Возможное выборочное акцентирование отдельных аспектов истории.
Доказательства паттернов искажения
Источник B · Эффект фрейминга
Available in two variants, GPT-5.4 Thinking and GPT-5.4 Pro, the model unifies reasoning, coding, and agentic workflows into a single release arriving just two days after GPT-5.3 Instant.