Across 11 comparable benchmarks, OpenAI o3 has better reported results on 11 and worse on 0.
Moderate evidence11 of 11 shared results countedHow we compare
Price basis: Standard rates. The price section below shows the applicable periods.
| Fact | OpenAI o3 | GPT-4.1 mini |
|---|---|---|
| Context window | 200K tokens | 1M tokens |
| Input / 1M | $2.00 | $0.400 |
| Output / 1M | $8.00 | $1.60 |
| Blended (4:1) | $2.12 | $0.424 |
| Status | Deprecated | Deprecated |
| Released | April 16, 2025 | April 14, 2025 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 9 shared benchmarks grouped by capability on a 0–100 scale. How averages are built
OpenAI o3 ahead on 3 of 3 counted
OpenAI o3 ahead on 3 of 3 counted
OpenAI o3 ahead on 1 of 1 counted
OpenAI o3 ahead on 1 of 1 counted
OpenAI o3 ahead on 1 of 1 counted
OpenAI o3 ahead on 1 of 1 counted
No fixed-scale average — see the rows in Benchmark results
No shared comparable results for Abstract reasoning, Computer & tool use, Research & search, Cybersecurity.
Published scores, with matching modes where available. Gaps show OpenAI o3 relative to GPT-4.1 mini. Reading the results
| Benchmark | OpenAI o3 | GPT-4.1 mini | Gap |
|---|---|---|---|
| Aider-PolyglotAgentic Development · Accuracy | 81.3Thinking High · no tools | 32.4≠Normal · no tools | +48.9 pts#4 vs #32 of 47 |
| SWE-bench VerifiedRepository Engineering · Accuracy | 69.1Thinking · no tools | 23.6≠Normal · no tools | +45.5 pts#65 vs #106 of 108 |
| Terminal Bench HardAgentic Development · Accuracy | 37.1Thinking · tools | 7.6≠Normal · tools | +29.5 pts#43 vs #122 of 157 |
| AIME 2024Mathematics · Accuracy | 91.6Normal · no tools | 49.6Normal · no tools | +42.0 pts#9 vs #43 of 54 |
| FrontierMathMathematics · Accuracy | 10.3Thinking Low · no tools | 4.5≠Normal · no tools | +5.8 pts#20 vs #30 of 48 |
| FrontierMath v2Mathematics · Accuracy (verification_code) | 33.3Thinking High · no tools | 6.7≠Reported best | +26.7 pts#69 vs #84 of 85 |
| GPQA DiamondScientific Reasoning · Accuracy | 83.3Thinking · no tools | 65.0≠Normal · no tools | +18.3 pts#94 vs #157 of 200 |
| Fiction.liveBenchLong Reasoning · 16k-token score (%) | 88.9Thinking Medium · no tools | 44.4≠Normal · no tools | +44.5 pts#6 vs #25 of 28 |
| MMMU-ProVisual Understanding · Accuracy | 76.4Reported best | 58.7≠Normal · no tools | +17.7 pts#56 vs #108 of 136 |
| Creative WritingWriting | 1675Normal · no tools | 1145Normal · no tools | 1.46×#37 vs #93 of 121 |
| ECICapability Indices · ECI score (capability index, higher is better) | 147Reported best | 135Reported best | 1.09×#62 vs #112 of 168 |
Agentic Development · Accuracy
Repository Engineering · Accuracy
Agentic Development · Accuracy
Mathematics · Accuracy
Mathematics · Accuracy
Mathematics · Accuracy (verification_code)
Scientific Reasoning · Accuracy
Long Reasoning · 16k-token score (%)
Visual Understanding · Accuracy
Writing
Capability Indices · ECI score (capability index, higher is better)
A dash means no result is available.
| Benchmark | OpenAI o3 | GPT-4.1 mini |
|---|---|---|
| AIME2025 | 88.9 | — |
| CritPt | 1.1 | — |
| FrontierMath - Tier 4 | 2.1 | — |
| HLE | 20.3 | — |
| IMO-ProofBench | 20.5 | — |
| IMO-ProofBench Advanced | 20.5 | — |
| MATH-500 | 99.2 | — |
| MMLU | — | 87.5 |
| MMLU-Pro | 85.6 | — |
| CodeClash | 1343 | — |
| GSO | 8.8 | — |
| IC SWE-Lancer (Diamond) | — | 33.0 |
| LiveCodeBench | 75.8 | — |
| Terminal-Bench 1.0 | 30.2 | — |
| WeirdML v2 | 52.4 | — |
| OSWorld-Verified | 23.0 | — |
| METR Time Horizons v1.1 | 91.3 | — |
| GeoBench ACW | 74.0 | — |
| MMMU | 82.9 | — |
| VPCT | 52.0 | — |
| SAGE | 41.8 | — |
| τ²-Bench | — | 53.0 |
| τ²-Bench - Telecom | — | 48.9 |
| τ³-Banking | — | 5.4 |
| ARC-AGI-1 | 60.8 | — |
| ARC-AGI-2 | 6.5 | — |
| SimpleBench | 53.1 | — |
| GAIA | 28.5 | — |
| SimpleQA | 49.4 | — |
| MedCode | 47.3 | — |
| MedScribe | 76.7 | — |
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix · 5% steps
Blended price = effective input rate at 90% cache hits × 80% + output rate × 20%
OpenAI o3 costs400%morethan GPT-4.1 mini per million tokens — 5× the price.Using standard rates at this token mix.
OpenAI o3: Standard pricing. OpenAI official pricing (developers.openai.com/api/docs/pricing) verified 2026-11-13; standard tier; Deprecated; shutdown 2026-12-11 (replacement: gpt-6.1-sol)
GPT-4.1 mini: Standard pricing. OpenAI official pricing (developers.openai.com/api/docs/pricing) verified 2026-11-13; standard tier; Deprecated; shutdown 2027-01-14 (replacement: gpt-6-luna)
| Price / 1M tokens | OpenAI o3 | GPT-4.1 mini | Ratio |
|---|---|---|---|
| Input | $2.00 | $0.40 | 5.00× |
| Output | $8.00 | $1.60 | 5.00× |
| Cached input | $0.50 | $0.10 | 5.00× |
| Cache write | — | — | — |
| Blended | $2.12 | $0.424 | 5.00× |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
Reported cost per task, paired with the score from the same run. Settings and sources may differ; costs do not affect the verdict.
| Attribute | OpenAI o3 | GPT-4.1 mini |
|---|---|---|
| Availability | Deprecated | Deprecated |
| Provider | OpenAI | OpenAI |
| Model type | Reasoning model | Chat model |
| First release | April 16, 2025 | April 14, 2025 |
| Context window | 200K tokens | 1M tokens |
| Max output | 100K tokens | 32K tokens |
| Architecture | Undisclosed | Undisclosed |
| Documented capabilities | Reasoning, Multilingual | Multilingual |
| Input modalities | Text, Image | Text, Image |
| Output modalities | Text | Text |
| Weights | Closed / API only | Closed / API only |
| Commercial use | Proprietary | Proprietary |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used OpenAI o3 or GPT-4.1 mini? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.