More exact GPT-5.4 tier for demanding professional reasoning and agent tasks
GPT-5.4 Pro by openai is offered by 20 providers on this page. Its official list price is $30.00 per 1M input tokens and $180 per 1M output tokens. The lowest paid channel is Ofox at $24.00 / $144 per 1M, about 1.9× below the list price. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it.
The context window is 1,050,000 tokens at the reference host, but hosts report different limits, from 1,000,000 to 1,050,000, so the usable window depends on the provider you pick. It supports reasoning and tool use. Accepted input modalities are Text, Image, and PDF. Its training knowledge cuts off at 2025-08-31.
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| Ofox | Gateway | $24.00 | $144 | — | — | 1,050,000 | 128,000 | |
| Poe | Gateway | $27.00 | $160 | — | — | 1,050,000 | 128,000 | |
| Kilo Gateway | Gateway | $30.00 | $180 | — | — | 1,050,000 | 128,000 | |
| OpenRouter | Gateway | $30.00 tiered >272K: $60.00 | $180 | — | — | 1,050,000 | 128,000 | |
| Azure | Cloud | $30.00 tiered >272K: $60.00 | $180 | — | — | 1,050,000 | 128,000 | |
| Azure Cognitive Services | Cloud | $30.00 tiered >272K: $60.00 | $180 | — | — | 1,050,000 | 128,000 | |
| Impossibl | Gateway | $30.00 tiered >272K: $60.00 | $180 | $3.00 | — | 1,050,000 | 128,000 | |
| Cloudflare AI Gateway | Cloud | $30.00 | $180 | — | — | 1,000,000 ⚠ | 128,000 | |
| Vercel AI Gateway | Cloud | $30.00 tiered >272K: $60.00 | $180 | — | — | 1,050,000 | 128,000 | |
| OrcaRouter | Gateway | $30.00 tiered >272K: $60.00 | $180 | — | — | 1,050,000 | 128,000 | |
| OpenAIOfficial | First-party | $30.00 tiered >272K: $60.00 | $180 | — | — | 1,050,000 | 128,000 | |
Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.
Interleaved thinking (reasoning between tool calls) is declared by 1 of 20 providers.
| Name | Conditions | Score | Metric | Source |
|---|---|---|---|---|
| GPQA Diamond | — | 94.4 | accuracy | Source ↗ |
| Humanity's Last Exam | variant: no tools | 42.7 | accuracy | Source ↗ |
| Humanity's Last Exam | variant: with tools | 58.7 | accuracy | Source ↗ |
| BrowseComp | — | 89.3 | accuracy | Source ↗ |
| GDPval | — | 82 | wins or ties | Source ↗ |
| FrontierMath | dataset: Tier 1-3 | 50 | accuracy | Source ↗ |
| FrontierMath | dataset: Tier 4 | 38 | accuracy | Source ↗ |
| ARC-AGI-1 | variant: Verified | 94.5 | accuracy | Source ↗ |
| ARC-AGI-2 | variant: Verified | 83.3 | accuracy | Source ↗ |
| FinanceAgent | v1.1 | 61.5 | accuracy | Source ↗ |
| GeneBench | — | 25.6 | accuracy | Source ↗ |
The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.