GPT-5.4 model page
Source of record for GPT-5.4 context window, max output tokens, and built-in tools including web search, file search, image generation, code interpreter, skills, and MCP.
OpenAI limits detail
This page is a decision brief for teams choosing between GPT-5.4 and GPT-5-mini under real context and tool pressure. It keeps the currently published context window, max output tokens, and built-in tool coverage in one source-linked read.
Why now
This is the part of the OpenAI decision that usually gets compressed into a single line in a spreadsheet. Context and tool support decide whether a cheaper row is actually deployable.
Official sources
This source set stays narrow on purpose so a model-fit decision can be audited without falling back to guesswork.
Source of record for GPT-5.4 context window, max output tokens, and built-in tools including web search, file search, image generation, code interpreter, skills, and MCP.
Source of record for GPT-5-mini context window, max output tokens, and its narrower built-in tool set.
Side-by-side fit
This table stays close to the current OpenAI model pages and avoids guessing beyond the documented support surface.
| Dimension | GPT-5.4 | GPT-5-mini | Why it matters | Sources |
|---|---|---|---|---|
| Context window | 1,048,576 tokens | 400,000 tokens | This is the biggest published fit gap between the two rows and often the main reason the cheaper option cannot simply replace the flagship path. | |
| Max output tokens | 128,000 tokens | 128,000 tokens | The output cap is currently not the differentiator here, so teams should not over-weight it in the swap decision. | |
| Built-in tools | Functions, web search, file search, image generation, code interpreter, skills, MCP | Functions, web search, file search, MCP | If the workflow needs image generation, code interpreter, or skills on the model path, GPT-5-mini is not a straight drop-in replacement. |
Decision signals
Use these signals when deciding whether to keep GPT-5.4, move down to GPT-5-mini, or split the workload between them.
If the workflow regularly approaches or exceeds the 400,000-token mark, GPT-5-mini should not be priced as if it were a full substitute for GPT-5.4.
A team can save money on the base model row and still lose the workflow if it forgets that GPT-5-mini does not currently list image generation, code interpreter, or skills on the same page.
Because both pages currently publish the same max output token limit, the choice should be led by context and tools instead of output ceiling alone.
Verify before switching
These are the minimum checks to run before replacing GPT-5.4 with GPT-5-mini in production.
Use production-like traces or representative prompts to confirm whether the workflow stays safely inside GPT-5-mini's published 400,000-token context ceiling instead of benchmarking on a simplified short-turn sample.
Before switching, verify whether the workflow depends on image generation, code interpreter, or skills, because those are currently listed on GPT-5.4 but not on GPT-5-mini.
If only the long-context or tool-heavy turns need GPT-5.4, the correct move may be workload routing instead of a full one-model swap.
Sources
Continue the site
Use the groups below to move laterally through the decision, not back out into another doc hunt.
Related pages
Stay in the same decision neighborhood instead of backing out to search.
Compare pages
Open the pages that turn this topic into a side-by-side decision.
Replacement pages
Use the likely substitutes, migration targets, or fallback choices as the next click.
Source category pages
Trace the source families behind this page instead of opening random docs in isolation.