vs
0:5
overall capability wins (2 ties across 7 scored). o4-mini leads overall.
- Reasoning0:2
- Coding0:2
- Arena0:1
- Speed2:0
Verdict
o4-mini leads coding (widest gap: +6.6 on SciCode); o4-mini ranks higher on LMArena (+43 Elo).
- o4-mini leads coding (widest gap: +6.6 on SciCode)
- o4-mini ranks higher on LMArena (+43 Elo)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- o3-mini
- $2.20
- o4-mini
- $2.20
Price per 1M tokens
Full comparison
Organization
OpenAI
OpenAI
Family
GPT-4 / o-series
GPT-4 / o-series
License
Proprietary
Proprietary
Open weights
No
No
Release
Jan 31, 2025
Apr 16, 2025
Knowledge cutoff
-
-
API / provider
OpenAI
OpenAI
Modalities
text → text
text, image → text
Specs
Context window
200K
200K
Max output
100K
100K
Parameters
—
—
Pricing
Input $/1M
$1.10
$1.10
Output $/1M
$4.40
$4.40
Blended $/1M (3∶1)
$1.93
$1.93
Speed
tok/s
100
90
TTFT (s)
0.6
0.7
Reasoning
MMLU-Pro
79.1
—
GPQA Diamond
74.8
78.4
Humanity's Last Exam
7.9
16.5
AIME 2025
—
92.7
MATH-500
97.3
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
—
24.8
AA-LCR v1.1
—
61.0
CritPt
—
0.6
MMMU-Pro
—
69.2
IFBench
—
68.7
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
45.0
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
71.7
—
Terminal-Bench 2.1
—
—
Aider Polyglot
53.8
58.2
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
39.9
46.5
CursorBench
—
—
SWE-Rebench
—
27.5
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,348
1,391
| Metric | o3-mini | o4-mini | Delta |
|---|---|---|---|
| Identity | |||
| Organization | OpenAI | OpenAI | — |
| Family | GPT-4 / o-series | GPT-4 / o-series | — |
| License | Proprietary | Proprietary | — |
| Open weights | No | No | — |
| Release | Jan 31, 2025 | Apr 16, 2025 | — |
| Knowledge cutoff | - | - | — |
| API / provider | OpenAI | OpenAI | — |
| Modalities | text → text | text, image → text | — |
| Specs | |||
| Context window | 200K | 200K | tie |
| Max output | 100K | 100K | tie |
| Parameters | — | — | — |
| Pricing | |||
| Input $/1M | $1.10 | $1.10 | tie |
| Output $/1M | $4.40 | $4.40 | tie |
| Blended $/1M (3∶1) | $1.93 | $1.93 | tie |
| Speed | |||
| tok/s | 100 | 90 | A +10 |
| TTFT (s) | 0.6 | 0.7 | A +0.10 |
| Reasoning | |||
| MMLU-Pro | 79.1 | — | — |
| GPQA Diamond | 74.8 | 78.4 | B +3.6 pts |
| Humanity's Last Exam | 7.9 | 16.5 | B +8.6 pts |
| AIME 2025 | — | 92.7 | — |
| MATH-500 | 97.3 | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | — | 24.8 | — |
| AA-LCR v1.1 | — | 61.0 | — |
| CritPt | — | 0.6 | — |
| MMMU-Pro | — | 69.2 | — |
| IFBench | — | 68.7 | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | 45.0 | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | 71.7 | — | — |
| Terminal-Bench 2.1 | — | — | — |
| Aider Polyglot | 53.8 | 58.2 | B +4.4 pts |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 39.9 | 46.5 | B +6.6 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | 27.5 | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,348 | 1,391 | B +43 Elo |
o3-mini
- Context
- 200K / 100K out
- Parameters
- —
- Price
- $1.10 / $4.40
- Speed
- 100 tok/s · 0.6s TTFT
- Modalities
- text → text
- License
- Proprietary
Earlier o-series mini - still useful for math-heavy workloads on a budget.
o4-mini
- Context
- 200K / 100K out
- Parameters
- —
- Price
- $1.10 / $4.40
- Speed
- 90 tok/s · 0.7s TTFT
- Modalities
- text, image → text
- License
- Proprietary
Cost-efficient o-series reasoning model for STEM and coding with lower latency than o3.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.