- Start
- Mar 5, 202695% CONFIDENCEfrom help.openai.com
GPT-5.4 Thinking Comes to ChatGPT
- GPT-5.4 Thinking arrived in ChatGPT on March 5, 2026, combining recent reasoning, coding and agentic advances with GPT-5.3-Codex coding[1][3]
- GPT-5.4 is available in ChatGPT (as GPT-5.4 Thinking), the API and Codex, with GPT-5.4 Pro in ChatGPT and the API, rolling out gradually[3]
- GPT-5.1 was retired on March 11, 2026 with conversations moving to GPT-5.3 Instant, GPT-5.4 Thinking or Pro, and GPT-5.4 mini followed on March 18[1]
Notable features
- GPT-5.4 is OpenAI's flagship model for complex professional work; in the API it takes text and image input, returns up to 128,000 output tokens, has an August 31, 2025 knowledge cutoff and supports reasoning effort from none (default) to xhigh, at $2.50 per million input and $15 per million output tokens, with prompts over 272K input tokens billed at 2x input and 1.5x output[2]
- It can give an upfront plan of its thinking and has improved deep web research and context management[1]
- Up to 1M tokens of context in Codex and the API on an opt-in basis[3]
- Native computer use scores 75.0% on OSWorld-Verified; GDPval matches or exceeds professionals in 83.0% of comparisons (GPT-5.2: 70.9%); internal spreadsheet benchmark 87.3% vs 68.4%[3]
- Tool search cut token usage 47% on 250 MCP Atlas tasks at the same accuracy[3]
References 391% CONFIDENCE
The first entry is always the pin's source. Overall confidence is a weighted average of how firmly each reference supports the start and end times used above; a reference counts half as much for every 180 days older than the newest.
- [1]95%OpenAI Model Release Notes: GPT-5.4 Thinking in ChatGPT (March 5, 2026)help.openai.com· Posted Sep 19, 2026· Starts Mar 5, 2026· 40% of score
OpenAI's[2][3] own model release notes date this entry March 5, 2026
- [2]90%GPT-5.4 Model | OpenAI APIdevelopers.openai.com· Added Sep 22, 2026· 41% of score
OpenAI's[1][3] API model page for GPT-5.4 calls it "a flagship model for complex professional work" and lists a 1,050,000 context window, 128,000 max output tokens, $2.50/$15 per 1M tokens and 2x/1.5x pricing for prompts over 272K tokens.
- [3]85%GPT-5.4 Pro and Thinking are here!community.openai.com· Published Mar 5, 2026· Starts Mar 5, 2026· 19% of score
OpenAI[1][2] staff announcement in the developer community, posted the day GPT-5.4 Thinking launched.
Suggest a correction
Something missing or wrong? Say it in your own words: a link that backs this pin up, a different start or end date and why, or a fact it lacks or gets wrong. The AI checks it against this pin's sources, searches for better ones, and adds any page that backs you up. The pin's own sources still count most.