GPT-5

OpenAI's landmark August 2025 flagship: strong reasoning at a low price

Context window

400,000 tokens (128,000 max output)

Input / 1M tokens

$1.25

Output / 1M tokens

$10.00

Provider

OpenAI

Cached input $0.125 per 1M. OpenAI recommends using the newer GPT-5.5 for new implementations, but GPT-5 remains available at this pricing. · Data verified 2026-07-02

GPT-5 is OpenAI's flagship model launched August 7, 2025, notable for combining frontier-level reasoning and coding with aggressively low API pricing ($1.25 input / $10 output per 1M). It has a 400,000-token context window with up to 128,000 output tokens, accepts text and image input, and has a knowledge cutoff of September 30, 2024. At launch it set state-of-the-art results including 74.9% on SWE-bench Verified and 94.6% on AIME 2025 (without tools), with a GPT-5 pro configuration reaching 88.4% on GPQA. It remains available, though OpenAI now recommends GPT-5.5 for new work.

Capability index

Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.

Coding

Reasoning

Math

Multimodal

Long context

Speed

Cost efficiency

How to access it

Available in the OpenAI API via model id 'gpt-5' (snapshot 'gpt-5-2025-08-07') and in ChatGPT. OpenAI recommends GPT-5.5 for new projects but GPT-5 is still fully supported.

Get access →Documentation →

Strengths

✓Very low price for a frontier-class model ($1.25/$10 per 1M)
✓Strong coding (SWE-bench Verified 74.9%) and math (AIME 2025 94.6% without tools)
✓Large 400K-token context window
✓Multimodal text + image input
✓Widely supported and well-documented

Best for developers who...

Budget-friendly frontier reasoning and codingMath and STEM problem solvingExisting GPT-5-based production systems

When to choose it (and when not to)

Reach for GPT-5 when...

→You want strong reasoning at the lowest frontier-model price
→Existing GPT-5-based systems that are already tuned and validated
→General coding, math, and reasoning workloads on a budget

Look elsewhere if...

✕New projects where OpenAI recommends the newer GPT-5.5
✕Tasks needing the largest context window (GPT-5.4/5.5 offer ~1.05M vs 400K)
✕Workloads requiring the latest knowledge (cutoff is September 30, 2024)

How to use it

›Disable or lower reasoning effort for simple tasks to save cost and latency
›Use cached input ($0.125/1M) for repeated system prompts and large prefixes
›Pin snapshot 'gpt-5-2025-08-07' for reproducibility
›For newer knowledge or bigger context, prefer GPT-5.5

Quickstart

Python

from openai import OpenAI

client = OpenAI()

response = client.responses.create(
    model="gpt-5",
    input="Write a Python function to parse ISO 8601 durations.",
)
print(response.output_text)

Snapshot 'gpt-5-2025-08-07' pins behavior. OpenAI recommends GPT-5.5 for new implementations.

API model id: gpt-5

Benchmarks

Benchmark	Score	Notes
SWE-bench Verified	74.9%	Real-world software engineering, per OpenAI's GPT-5 launch.
AIME 2025	94.6%	Competition math without tools, per OpenAI.
GPQA (GPT-5 pro)	88.4%	Graduate-level science with extended reasoning (pro configuration), without tools.

Source: OpenAI - Introducing GPT-5