OpenAI

OpenAI: GPT-4o-mini (2024-07-18)

Type
language model
Context
128K tokens
Max output
16K
Released
18 July 2024
Knowledge cutoff
October 2023
API string
openai/gpt-4o-mini-2024-07-18

Prices by provider

Provider Input, $ per 1M Output, $ per 1M In our data since
Kilo Code $0.15 $0.6 31 Jul 2026
OpenAI $0.15 $0.6 31 Jul 2026
OpenRouter $0.15 $0.6 31 Jul 2026
Microsoft Foundry $0.165 $0.66 31 Jul 2026
Venice AI $0.188 $0.75 1 Aug 2026

Only the base tier and only per-token prices are shown. Batch, discounted and cached rates, as well as prices per image or per second of video, do not go into this table: they cannot stand in the same column as a price per million tokens.

The date in the last column is the day this price first entered our collection. The price may well be older: before that day we simply were not recording it. It has not changed since — otherwise a new row with a new date would stand in its place.

Measurement results

Task set Result Run conditions Measured by
Arena Score in English 1,300.59 1,296–1,305
Arena Score in French 1,295.68 1,271–1,320
Arena Score, programming 1,290.12 1,284–1,297
Arena Score, long queries 1,288.94 1,282–1,296
Arena Score, overall 1,286.57 1,283–1,290
Arena Score, multi-turn dialogue 1,285.04 1,278–1,292
Arena Score in Spanish 1,278.34 1,255–1,302
Arena Score in Russian 1,274.29 1,267–1,282
Arena Score, creative writing 1,268.26 1,261–1,275
Arena Score, mathematics 1,267.33 1,260–1,274
Arena Score, hard prompts 1,267.19 1,262–1,272
Arena Score, instruction following 1,258.81 1,254–1,264
Arena Score, expert questions 1,233.87 1,223–1,245
Arena Score, working with images 1,065.76 1,058–1,074
Creative writing (Lech Mazur’s evaluation) 67.2 % lechmazur/writing Github repository
GeoBench — locating a place from a photograph 64 % GeoBench leaderboard
MATH, difficulty level five 52.63 % · with a tuned harness Epoch evaluations
BALROG — game environments 17.4 % Balrog Leaderboard
GPQA Diamond — graduate-level questions 16.96 % · with a tuned harness Epoch evaluations
WeirdML — unusual machine learning tasks 11.76 % WeirdML Leaderboard
Mock AIME 2024–2025 — olympiad problems 6.85 % · with a tuned harness Epoch evaluations
Aider Polyglot — code edits in six languages 3.6 % · with a tuned harness Aider LLM Leaderboards
ARC-AGI-2 0 %
SimpleBench — trick questions 0 % SimpleBench Leaderboard
Chess puzzles 0 % Epoch evaluations