gpt-4o-realtime-preview is a frontier-tier language model from OpenAI. gpt-4o-realtime-preview is a proprietary model available exclusively through OpenAI's API. GPT-4 is OpenAI's flagship model family, known for strong instruction following, broad knowledge, and reliable function calling — the most widely integrated LLM family in production applications.
gpt-4o-realtime-preview is priced at $5/1M input tokens and $20/1M output tokens. Output tokens are 4× more expensive than input tokens — this ratio reflects the sequential nature of autoregressive generation, which requires a separate forward pass per output token versus parallel processing of all input tokens simultaneously. At a typical usage pattern of 1,000 input tokens and 500 output tokens per call, gpt-4o-realtime-preview costs approximately $15.00 per 1,000 API calls. OpenAI does not publish a prompt caching rate for this model, so repeated long contexts are billed at the standard $5/1M input rate.
Its 128K-token context window handles most long-document tasks — roughly 75,000 words or a full novel — without requiring chunking or retrieval augmentation. Median latency (P50 time-to-first-token) is approximately 300ms — suitable for interactive applications where sub-second response is acceptable. Maximum output length is 4K tokens — adequate for most generation tasks including essays, code functions, and detailed responses.
gpt-4o-realtime-preview is best suited to voice, realtime, chat. Supported capabilities include function calling / tool use for agentic workflows, streaming token output for real-time UIs. Function calling support makes it suitable for building AI agents, structured data extraction pipelines, and any workflow requiring the model to invoke external APIs or return typed JSON. When evaluating this model for a specific workload, the most important factors are: context window (does your input fit?), output price (how long are your generations?), and latency (is this interactive or batch?). The pricing table above shows how gpt-4o-realtime-preview compares to similar models across these dimensions.
gpt-4o-realtime-preview sits in the frontier tier — the highest-capability models available, used when quality is the primary constraint and cost is secondary. Unlike open-weights alternatives, gpt-4o-realtime-preview is only available via OpenAI's API — there is no self-hosted option, which simplifies deployment but creates vendor dependency. When comparing gpt-4o-realtime-preview to alternatives, use the similar models section on this page to find models at comparable price points with different capability tradeoffs. The intelligence score (0/100) reflects aggregate benchmark performance — a useful proxy for general capability, though real-world performance varies significantly by task type. Always test with your specific prompts and use case before committing to a model for production.
Composite quality score based on reasoning, coding, and instruction-following benchmarks.
Input price vs 135 frontier models tracked. Avg: $5.69/1M.
| Model ID | gpt-4o-realtime-preview |
| Provider | OpenAI |
| Model Family | GPT-4o |
| Tier | Frontier |
| Context Window | 128,000 tokens (128K) |
| Max Output Tokens | 4,096 tokens |
| Input Price | $5 / 1M tokens |
| Output Price | $20 / 1M tokens |
| Cached Input Price | Not available |
| Throughput | — |
| Latency (P50) | 300ms |
| Intelligence Score | 0/100 |
| LiveBench Score | — |
| Cost per Task (LiveBench) | — |
| Data Confidence | — |
| HLE Score | — |
| GPQA Diamond | — |
| SWE-bench Score | — |
| OSWorld Score | — |
| BrowseComp Score | — |
| Terminal-Bench Score | — |
| AutoBench Score | — |
| Parameters | — |
| Vision Support | No |
| Function Calling | Yes |
| Streaming | Yes |
| Reasoning Model | No |
| Open Weights | No |
| Data Source | Static (verified) |
| Last Updated | 2026-07-01 |
Similar Models by Intelligence Score
Self-hosting gpt-4o-realtime-preview? These GPUs are well-matched for this model's compute and memory requirements.