New

Work in progress: Agents Directory has just launched. Stay tuned, more content is on the way.

GPT-5.4 Mini

GPT-5.4 mini brings the core capabilities of GPT-5.4 to a faster, more efficient model optimized for high-throughput workloads. It supports text and image inputs with strong performance across reasoning, coding,...

Capabilities:

Input

Text input
Image input (vision)
File input (PDF)
Audio input
Video input

Output

Text output
Image output
Audio output

Benchmark scores:

BenchmarkRankScore

FrontierCode Main
Rank#5 of 12
Score17.8%
FrontierCode Diamond
Rank#5 of 11
Score4.6%

GPT-5.4 Mini on FrontierCode Main

17.8%

Cognition's test of whether a model writes code maintainers would actually merge, not just code that passes tests. Main is the 100 hardest of 150 tasks. Higher is better.

GPT-5.4 Mini scores 17.8%, #5 of 12 tracked models.
Claude Fable 5 leads at 46.3%. The group median is 16.35%.

vs. top models

Full interactive leaderboard on our FrontierCode Main page.

Source: cognition.ai · Updated June 2026

GPT-5.4 Mini on FrontierCode Diamond

4.6%

The 50 hardest FrontierCode tasks: the toughest production-code problems, graded on whether maintainers would merge the patch. Scores stay low by design. Higher is better.

GPT-5.4 Mini scores 4.6%, #5 of 11 tracked models.
Claude Opus 4.8 leads at 13.4%. The group median is 3.8%.

vs. top models

Full interactive leaderboard on our FrontierCode Diamond page.

Source: cognition.ai · Updated June 2026

Pricing & availability:

OpenRouter
$0.75 / $4.5 per M

Appears in rankings:

Best models for Codex

Sources:

OpenRouter

Share:

Details:

Provider
OpenAI
Context window
400K
Input price
$0.75/M
Output price
$4.5/M
Open weights
No
Knowledge cutoff
Aug 2025
Released
Mar 2026