GPT-5.4 minivsGPT-5-mini
Across 9 shared benchmarks, GPT-5.4 mini leads overall: GPT-5.4 mini wins 5, GPT-5-mini wins 4, with 0 ties and an average score difference of +42.09.
GPT-5.4 mini
OpenAI · 2026-03-17 · Reasoning model
GPT-5-mini
OpenAI · 2025-08-07 · Foundation model
GPT-5.4 mini5 wins(56%)(44%)4 winsGPT-5-mini
Benchmark scores
Grouped by capability, sorted by largest gap within each. 9 shared benchmarks.
Agent Level Benchmark
Even 2/2| Benchmark | GPT-5.4 mini | GPT-5-mini | Diff |
|---|---|---|---|
| τ²-Bench - Telecom | 23.40234 / 264Normal (With Tools) | 31.90206 / 264Normal (With Tools) | -8.50 |
| Terminal Bench Hard | 18.20154 / 244Normal (With Tools) | 14.40170 / 244Normal (With Tools) | +3.80 |
General Knowledge
Even 2/2| Benchmark | GPT-5.4 mini | GPT-5-mini | Diff |
|---|---|---|---|
| LiveBench | 36.95114 / 117Normal (No Tools) | 61.0168 / 117Normal (No Tools) | -24.06 |
| HLE | 5.90462 / 563Normal (No Tools) · Text only | 5.10478 / 563Normal (No Tools) · Text only | +0.80 |
General Evaluation
GPT-5.4 mini 1/1| Benchmark | GPT-5.4 mini | GPT-5-mini | Diff |
|---|---|---|---|
| GPQA Diamond | 64.14362 / 462Normal (No Tools) | 0461 / 462Normal (No Tools) | +64.14 |
Instruction Following
GPT-5-mini 1/1| Benchmark | GPT-5.4 mini | GPT-5-mini | Diff |
|---|---|---|---|
| IF Bench | 38.80225 / 282Normal (No Tools) | 45.60178 / 282Normal (No Tools) | -6.80 |
Math and Reasoning
GPT-5-mini 1/1| Benchmark | GPT-5.4 mini | GPT-5-mini | Diff |
|---|---|---|---|
| FrontierMath - Tier 4 | 2.1056 / 80Thinking High (No Tools) | 6.3035 / 80Thinking High (No Tools) | -4.20 |
Multimodal Understanding
GPT-5.4 mini 1/1| Benchmark | GPT-5.4 mini | GPT-5-mini | Diff |
|---|---|---|---|
| MMMU-Pro | 60.50172 / 227Normal (No Tools) | 58.40180 / 227Normal (No Tools) | +2.10 |
Writing and Creative Capabilities
GPT-5.4 mini 1/1| Benchmark | GPT-5.4 mini | GPT-5-mini | Diff |
|---|---|---|---|
| Creative Writing | 1,66234 / 106Normal (No Tools) | 1,31072 / 106Normal (No Tools) | +351.50 |
Specs
| Field | GPT-5.4 mini | GPT-5-mini |
|---|---|---|
| Publisher | OpenAI | OpenAI |
| Release date | 2026-03-17 | 2025-08-07 |
| Model type | Reasoning model | Foundation model |
| Architecture | Dense | Dense |
| Parameters | Not available | Not available |
| Context length | 400K | 400K |
| Max output | 128K | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | GPT-5.4 mini | GPT-5-mini |
|---|---|---|
| Text input | $0.75 / 1M tokens | $0.25 / 1M tokens |
| Text output | $4.5 / 1M tokens | $2 / 1M tokens |
| Cache read | $0.075 / 1M tokens | $0.025 / 1M tokens |
| Cache write | Not public | $0 / 1M tokens |
Summary
- GPT-5.4 minileads in:General Evaluation (1/1), Multimodal Understanding (1/1), Writing and Creative Capabilities (1/1)
- GPT-5-minileads in:Instruction Following (1/1), Math and Reasoning (1/1)
- Tied in:Agent Level Benchmark, General Knowledge
On average across the 9 shared benchmarks, GPT-5.4 mini scores 42.09 higher.
Largest single-benchmark gap: Creative Writing — GPT-5.4 mini 1,662 vs GPT-5-mini 1,310 (+351.50).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.