OpenAI

o4 Mini 2025-04-16

Type
language model
Context
200K tokens
Max output
100K
API string
azure/o4-mini-2025-04-16

Prices by provider

Provider Input, $ per 1M Output, $ per 1M In our data since
Microsoft Foundry azure/o4-mini-2025-04-16 $1.1 $4.4 31 Jul 2026
OpenAI $1.1 $4.4 31 Jul 2026
Microsoft Foundry azure/us/o4-mini-2025-04-16 $1.21 $4.84 31 Jul 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.

A single provider appears several times if it sells this model under different identifiers and at different prices — for example, at different weight precision. The identifier is shown next to the name. Identical offers that arrived from two sources under different spellings are merged into one row.

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, mathematics 1,386.94 1,376–1,398
Arena Score, programming 1,368.52 1,362–1,375
Arena Score in English 1,366.19 1,361–1,371
Arena Score in French 1,359.86 1,334–1,385
Arena Score, overall 1,352.74 1,349–1,357
Arena Score, hard prompts 1,350.98 1,346–1,356
Arena Score, multi-turn dialogue 1,349.24 1,342–1,356
Arena Score in Spanish 1,345.84 1,325–1,366
Arena Score, expert questions 1,343.59 1,331–1,356
Arena Score in Russian 1,335.24 1,325–1,346
Arena Score, instruction following 1,321.33 1,315–1,327
Arena Score, long queries 1,314.94 1,308–1,322
Arena Score, creative writing 1,295.6 1,287–1,304
Arena Score, text recognition in images 1,197.64 1,189–1,206
Arena Score, understanding diagrams 1,197.47 1,185–1,210
Arena Score, working with images 1,194.83 1,188–1,202
MATH, difficulty level five 97.83 % effort: high · with a tuned harness Epoch evaluations
Mock AIME 2024–2025 — olympiad problems 81.65 % effort: low · with a tuned harness Epoch evaluations
Fiction.LiveBench — holding a long context 77.8 % effort: medium Fiction.live leaderboard
Creative writing (Lech Mazur’s evaluation) 75 % effort: medium lechmazur/writing Github repository
GPQA Diamond — graduate-level questions 72.81 % effort: low · with a tuned harness Epoch evaluations
Aider Polyglot — code edits in six languages 72 % effort: high · with a tuned harness Aider LLM Leaderboards
GeoBench — locating a place from a photograph 64 % effort: high GeoBench leaderboard
CadEval — building CAD models with code 62 % effort: medium CadEval Dashboard
ARC-AGI — generalising to unseen patterns 58.7 % effort: high · with a tuned harness ARC Prize Leaderboard
WeirdML — unusual machine learning tasks 52.56 % effort: high WeirdML Leaderboard
FrontierMath, levels 1–3 36.14 % effort: high Epoch evaluations
SimpleBench — trick questions 26.44 % effort: high SimpleBench Leaderboard
GDPval — tasks from real occupations 25.3 % effort: high
SimpleQA Verified — factual accuracy 23.9 % effort: low Epoch evaluations
Chess puzzles 22.14 % effort: low Epoch evaluations
Humanity’s Last Exam — expert-level questions 13.95 % effort: high
ARC-AGI-2 6.11 % effort: high
FrontierMath, level 4 — research-grade problems 4.88 % effort: high Epoch evaluations
GSO-Bench — code optimisation 3.6 % effort: high GSO Leaderboard
CritPt — physics problems 0.6 % effort: high