GPT-4o
OpenAI's versatile, fast multimodal workhorse (text + image)
OpenAI has replaced this model with GPT-6 Luna.
Context window
128,000 tokens (16,384 max output)
Input / 1M tokens
$2.50
Output / 1M tokens
$10.00
Provider
OpenAI
Cached input $1.25 per 1M. OpenAI notified developers on October 14, 2026 that the gpt-4o-2024-08-06 and gpt-4o-2024-11-20 snapshots (and GPT-4o mini) will be removed from the API on January 14, 2027, per its deprecations page; the recommended replacement is gpt-6-luna. The original gpt-4o-2024-05-13 snapshot was shut down on October 23, 2026. Fine-tuned GPT-4o models follow the same schedule. · Data verified 2026-11-08
GPT-4o ('omni') is OpenAI's multimodal model launched May 13, 2024, designed for fast, cost-effective general use. It accepts text and image input and outputs text, with a 128,000-token context window and up to 16,384 output tokens, and a knowledge cutoff of October 1, 2023. Priced at $2.50 input / $10 output per 1M, it delivers solid results across knowledge (MMLU 88.7), coding (HumanEval 90.2), and math. OpenAI deprecated its API snapshots on October 14, 2026, with removal on January 14, 2027 and GPT-6 Luna as the recommended replacement.
Capability index
Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.
How to access it
Available in the OpenAI API via model id 'gpt-4o' (snapshots 'gpt-4o-2024-08-06' and 'gpt-4o-2024-11-20') until OpenAI removes both snapshots on January 14, 2027. OpenAI recommends migrating to gpt-6-luna.
Strengths
- ✓Fast and inexpensive for a capable general-purpose model
- ✓Multimodal text + image input
- ✓Strong everyday coding and knowledge performance (HumanEval 90.2, MMLU 88.7)
- ✓Mature, extremely well-supported across tooling and integrations
Best for developers who...
When to choose it (and when not to)
Reach for GPT-4o when...
- →Everyday assistant, drafting, summarization, and classification tasks
- →Latency- and cost-sensitive applications at scale
- →Multimodal tasks needing image understanding with fast responses
Look elsewhere if...
- ✕New projects: OpenAI removes the GPT-4o API snapshots on January 14, 2027 (use gpt-6-luna)
- ✕Hard reasoning, competition math, or complex agentic coding (use GPT-5-series)
- ✕Very long documents beyond 128K tokens (GPT-5.4/5.5 offer ~1.05M)
- ✕Tasks needing knowledge after October 2023
How to use it
- ›Be explicit and structured; GPT-4o follows direct instructions well without heavy reasoning prompts
- ›Use cached input ($1.25/1M) for repeated system prompts to cut costs
- ›For image inputs, ask targeted questions about the image rather than open-ended prompts
- ›Escalate to a GPT-5-series model when a task needs deeper reasoning or bigger context
Quickstart
Pythonfrom openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": "Summarize this email in one sentence."}],
)
print(response.choices[0].message.content)Pin snapshot 'gpt-4o-2024-11-20' for the latest stable behavior. gpt-4o supports text and image input, text output.
API model id: gpt-4o
Benchmarks
| Benchmark | Score | Notes |
|---|---|---|
| MMLU | 88.7% | General knowledge, per OpenAI's GPT-4o evaluations. |
| HumanEval | 90.2% | Code generation, per OpenAI. |
| MATH | 76.6% | Competition math, per OpenAI's GPT-4o evaluations. |
Source: OpenAI - Hello GPT-4o
Tools powered by GPT-4o
Compare GPT-4o
Compare GPT-4o with any other model
Build a comparison →All model comparisons →