← All models

GPT-4o

Language Deprecated Full comparison ↗

openai/gpt-4o · by OpenAI · decoder-only-transformer

Pricing — 1 offering(s)

About this offering: OpenAI's GPT-4o (omni) multimodal model with a 128K-token context, at $2.50 input and $10 output per 1M tokens. OpenAI shuts down the original gpt-4o-2024-05-13 snapshot ($5 / $15) on 2026-10-23, with GPT-5.6 Sol as its replacement. On 2026-10-14 OpenAI announced that GPT-4o shuts down on 2027-01-14: the gpt-4o-2024-08-06 snapshot (the gpt-4o alias's default) and gpt-4o-2024-11-20 are on its shutdown list, with gpt-6-luna as the recommended replacement (OpenAI deprecations page, checked 2026-10-24). Price history here starts on 2026-06-15: OpenAI doesn't publish when the gpt-4o alias moved to today's price (it launched on the earlier 2024-05-13 snapshot), so earlier dates aren't shown rather than shown with an unsourced price.

Input tokens

  • $2.50 / 1M tokens (input) Current 2026-06-15 → 2027-01-13

Output tokens

  • $10.00 / 1M tokens (output) Current 2026-06-15 → 2027-01-13

Batch

Batch input tokens

  • $1.25 / 1M tokens (input) Current 2026-11-03 → 2027-01-13

Batch output tokens

  • $5.00 / 1M tokens (output) Current 2026-11-03 → 2027-01-13

Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.

See how GPT-4o fits into a cost-aware routing setup

See how →

Capability profile

How GPT-4o rates across core capability dimensions, with the task-level evidence behind each rating.

reasoning Strong
coding Moderate

SWE-bench Verified (500 human-verified GitHub issues; patch must pass the repo's own tests; no file-location hints): resolves 38.2% — fails roughly 3 issues in 5. Independent (official SWE-bench leaderboard; agentless 1.5, gpt-4o-2024-08-06).

tool use Strong
instruction following Strong
context window Moderate

128K context.

multilingual Strong
speed Moderate
cost efficiency Moderate
Benchmark Score Harness Source
swe-bench-verified ↗ 38.2% 2024-10 agentless 1.5 📊 leaderboard ↗
aider-polyglot ↗ 44.6% 2025-01 aider 📊 leaderboard ↗
bigcodebench ↗ 61.2% 2024-09 — 📊 leaderboard ↗

Operator guidance

Route here when GPT-4 capability is needed at a significantly lower price, or when multimodal input is required. Step up to o3 for deep reasoning tasks, or down to GPT-4o mini for high-volume cheap completions.

Use cases

Limitations

Citations