Current as of March 2026. GPT-5.4 adds 1.1M context to the 5.x line — a significant jump from the 272K window in 5.1 and 5.2. The output limit stays at 128K. If you’re running into context ceiling problems with 5.2, this is where you go next. The price reflects it: $2.50/$15 per million.
Specs
| Provider | OpenAI |
| Input cost | $2.50 / M tokens |
| Output cost | $15 / M tokens |
| Context window | 1.1M tokens |
| Max output | 128K tokens |
| Parameters | N/A |
| Features | function_calling, vision, reasoning |
What it’s good at
1.1M Context Window
This is the primary reason to use it over 5.2. Entire documentation sites, large monorepos, multi-document research sets — you can fit them without chunking.
Reasoning at Scale
Multi-step planning across a large context is where this model performs well. Cross-file dependency mapping, complex refactors, architectural analysis.
128K Output Ceiling
Same as 5.1/5.2. Generate large code files or long-form reports without hitting a wall mid-response.
Where it falls short
Output Cost
$15/M is steep. 6x the input rate. If you’re generating large outputs frequently, run the numbers before committing.
TTFT
Reasoning adds latency. Interactive applications will feel slow.
Best use cases with OpenClaw
- Full-Repository Refactoring — The context window is the differentiator here. Feed in the whole repo and get coherent cross-file changes back.
- Research Synthesis — Thousands of pages of PDFs, one prompt. The reasoning layer keeps the logic consistent at scale.
Not ideal for
- Simple Chat — Wasteful. Use GPT-4o-mini.
- High-Frequency Log Parsing — Latency and cost make this wrong for streaming or high-velocity data.
Run it through Haimaker
Skip juggling API keys. One Haimaker key gives you access to every model on the platform. Tell OpenClaw:
Add Haimaker as a custom provider to my OpenClaw config. Use these details:
- Provider name: haimaker
- Base URL: https://api.haimaker.ai/v1
- API key: [PASTE YOUR HAIMAKER API KEY HERE]
- API type: openai-completions
Add the auto-router model:
- haimaker/auto (reasoning: false, context: 128000, max tokens: 32000)
Create an alias "auto" for easy switching. Apply the config when done.
Or skip model selection entirely — Haimaker’s auto-router picks the best model for each task so you don’t have to.
OpenClaw setup
OpenClaw treats this as a first-class citizen. Export your OPENAI_API_KEY to your environment and you are ready to go with zero extra configuration.
export OPENAI_API_KEY="your-key-here"
That’s it. OpenClaw picks up OpenAI models automatically.
How it compares
- vs GPT-5.2 — 5.2 is cheaper ($14/M output) but tops out at 272K context. If you’re not hitting that ceiling, stick with 5.2.
- vs Gemini 1.5 Pro — Gemini has a 2M window and is cheaper on input. But GPT-5.4’s function calling holds up better in OpenClaw’s tool loops.
Bottom line
The right call when you’ve outgrown 272K context and need the reasoning quality of the 5.x line. Otherwise, 5.2 does the job for less.
For setup instructions, see our API key guide. For all available models, see the complete models guide.