Model comparison
GPT-5.2 vs GPT-5.4
Head-to-head evidence from 23 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Public leaderboard positions: GPT-5.2 #64 (Estimated); GPT-5.4 #8 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. GPT-5.2 and GPT-5.4 share 23 comparable benchmark results. 5 of 8 categories are comparable. 5 results are unique to GPT-5.2; 29 to GPT-5.4.
Updated July 20, 2026- Shared results
- 23
- GPT-5.2 only
- 5
- GPT-5.4 only
- 29
- Comparable categories
- 5 / 8
Pick GPT-5.4 if you want the stronger benchmark profile. GPT-5.2 only becomes the better choice if knowledge is the priority or you want the cheaper token bill.
Confidence note. This is a partial-evidence comparison with 23 shared benchmark results across 7 evidence categories; 5 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
GPT-5.4 is clearly ahead on the BenchAlign aggregate, 74.24 to 58.43. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
GPT-5.4's sharpest advantage is in agentic, where it averages 77.2 against 55.7. The single biggest benchmark swing on the page is OSWorld-Verified, 47.3% to 75%. GPT-5.2 does hit back in knowledge, so the answer changes if that is the part of the workload you care about most.
GPT-5.4 is also the more expensive model on tokens at $2.50 input / $15.00 output per 1M tokens, versus $1.75 input / $14.00 output per 1M tokens for GPT-5.2. GPT-5.4 gives you the larger context window at 1.05M, compared with 400K for GPT-5.2.
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | GPT-5.2 | Δ | GPT-5.4 |
|---|---|---|---|
| Knowledge | GPT-5.292.4 | Margin← 34.8 | GPT-5.457.6 |
| Agentic | GPT-5.255.7 | Margin→ 21.5 | GPT-5.477.2 |
| Coding | GPT-5.270.6 | Margin← 12.9 | GPT-5.457.7 |
| Math | GPT-5.235.2 | Margin→ 7.3 | GPT-5.442.5 |
| Multimodal | GPT-5.280.4 | Margin← 7.2 | GPT-5.473.2 |
| Reasoning | GPT-5.252.9 | MarginNo overlap | GPT-5.4Not measured |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
OSWorld-Verified
AgenticA 47.3%B 75%Winner: GPT-5.4Δ 27.7OSWorld-Verified: GPT-5.2 scored 47.3%; GPT-5.4 scored 75%. GPT-5.4 wins this benchmark. - Source ↗
BrowseComp
AgenticA 65.8%B 82.7%Winner: GPT-5.4Δ 16.9BrowseComp: GPT-5.2 scored 65.8%; GPT-5.4 scored 82.7%. GPT-5.4 wins this benchmark. - Source ↗
FrontierMath v2 (Tier 4)
MathA 18.800%B 27.100%Winner: GPT-5.4Δ 8.3FrontierMath v2 (Tier 4): GPT-5.2 scored 18.800%; GPT-5.4 scored 27.100%. GPT-5.4 wins this benchmark. - Source ↗
FrontierMath v2 (Tiers 1-3)
MathA 40.700%B 47.600%Winner: GPT-5.4Δ 6.9FrontierMath v2 (Tiers 1-3): GPT-5.2 scored 40.700%; GPT-5.4 scored 47.600%. GPT-5.4 wins this benchmark. - Source ↗
SWE-bench Pro
CodingA 55.6%B 57.7%Winner: GPT-5.4Δ 2.1SWE-bench Pro: GPT-5.2 scored 55.6%; GPT-5.4 scored 57.7%. GPT-5.4 wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | GPT-5.2 | GPT-5.4 | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | GPT-5.2$1.75 input / $14 output | GPT-5.4$2.5 input / $15 output | GPT-5.2 has the lower combined listed price. |
| Generation speedtokens per second | GPT-5.273 tok/s | GPT-5.474 tok/s | GPT-5.4 has the higher measured throughput. |
| First-answer latencyseconds to first token | GPT-5.2130.34 s | GPT-5.4151.79 s | GPT-5.2 reaches the first token sooner. |
| Context windowmaximum listed tokens | GPT-5.2400K | GPT-5.41.05M | GPT-5.4 lists the larger context window. |
Benchmark Deep Dive
AgenticGPT-5.4 wins17 benchmarks
| Benchmark | GPT-5.2 | GPT-5.4 | Result |
|---|---|---|---|
| BrowseCompSource | 65.8% | 82.7% | GPT-5.4 leads |
| OSWorld-VerifiedSource | 47.3% | 75% | GPT-5.4 leads |
| τ²-bench resultsSource | 84.8% | 87.1% | GPT-5.4 leads |
| Gert LabsSource | 46.54% | 64.89% | GPT-5.4 leads |
| JobBenchSource | 34.3% | 38.9% | GPT-5.4 leads |
| Terminal-Bench 2.0Source | — | 75.1% | Not comparable |
| CyberGymSource | — | 79.0% | Not comparable |
| MCP AtlasSource | — | 70.6% | Not comparable |
| ToolathlonSource | — | 54.6% | Not comparable |
| Claw-EvalSource | — | 60.3% | Not comparable |
| DeepSearchQASource | — | 73.6% | Not comparable |
| AA Agentic IndexSource | — | 41.1% | Not comparable |
| APEX-Agents-AASource | — | 33.3% | Not comparable |
| GDPval-AASource | — | 44.7% | Not comparable |
| GDPval-AASource | — | 1395 | Not comparable |
| ResearchClawBenchSource | — | 15.3% | Not comparable |
| ExploitGymSource | — | 6.0% | Not comparable |
CodingGPT-5.2 wins7 benchmarks
| Benchmark | GPT-5.2 | GPT-5.4 | Result |
|---|---|---|---|
| SWE-bench VerifiedSource | 80% | — | Not comparable |
| SWE-bench ProSource | 55.6% | 57.7% | GPT-5.4 leads |
| Vibe Code BenchSource | 53.50% | 67.42% | GPT-5.4 leads |
| AA-SciCodeSource | 52.1% | 56.6% | GPT-5.4 leads |
| LiveCodeBench ProSource | — | 87.5% | Not comparable |
| React Native EvalsSource | — | 85.3% | Not comparable |
| AA Coding IndexSource | — | 71.0% | Not comparable |
Reasoning3 benchmarks
KnowledgeGPT-5.2 wins13 benchmarks
| Benchmark | GPT-5.2 | GPT-5.4 | Result |
|---|---|---|---|
| GPQASource | 92.4% | 92.8% | GPT-5.4 leads |
| Artificial Analysis Intelligence IndexSource | 42.2% | 51.4% | GPT-5.4 leads |
| AA-GPQA DiamondSource | 90.3% | 92.0% | GPT-5.4 leads |
| AA-HLESource | 35.4% | 41.6% | GPT-5.4 leads |
| AA-Omniscience IndexSource | -1.0% | 5.7% | GPT-5.4 leads |
| AA-Omniscience AccuracySource | 43.8% | 50.0% | GPT-5.4 leads |
| AA-Omniscience Hallucination RateSource | 79.7% | 88.6% | GPT-5.2 leads |
| HLESource | — | 52.1% | Not comparable |
| HLE w/o toolsSource | — | 39.8% | Not comparable |
| GPQA-DSource | — | 92.8% | Not comparable |
| HealthBench HardSource | — | 40.1% | Not comparable |
| MedXpertQA (Text)Source | — | 59.6% | Not comparable |
| HealthBench ProfessionalSource | — | 48.1% | Not comparable |
MathGPT-5.4 wins3 benchmarks
MultimodalGPT-5.2 wins13 benchmarks
| Benchmark | GPT-5.2 | GPT-5.4 | Result |
|---|---|---|---|
| MMMU-ProSource | 79.5% | 81.2% | GPT-5.4 leads |
| MathVisionSource | 83.0% | — | Not comparable |
| CharXivSource | 82.1% | 82.8% | GPT-5.4 leads |
| V*Source | 75.9% | — | Not comparable |
| Design Arena WebsiteSource | 1227 | 1252 | GPT-5.4 leads |
| OfficeQA ProSource | — | 53.2% | Not comparable |
| MMMU-Pro w/ PythonSource | — | 82.1% | Not comparable |
| ERQASource | — | 65.4% | Not comparable |
| SimpleVQASource | — | 61.1% | Not comparable |
| ScreenSpot ProSource | — | 85.4% | Not comparable |
| ZeroBenchSource | — | 41.0% | Not comparable |
| MedXpertQA (MM)Source | — | 77.1% | Not comparable |
| AA-MMMU-ProSource | — | 78.4% | Not comparable |
Inst. Following1 benchmarks
| Benchmark | GPT-5.2 | GPT-5.4 | Result |
|---|---|---|---|
| AA-IFBenchSource | 75.4% | 73.9% | GPT-5.2 leads |
Frequently Asked Questions (6)
Which is better, GPT-5.2 or GPT-5.4?
GPT-5.4 is ahead on BenchLM's BenchAlign leaderboard, 74.24 to 58.43. The biggest single separator in this matchup is OSWorld-Verified, where the scores are 47.3% and 75%.
Which is better for knowledge tasks, GPT-5.2 or GPT-5.4?
GPT-5.2 has the edge for knowledge tasks in this comparison, averaging 92.4 versus 57.6. Inside this category, Artificial Analysis Intelligence Index is the benchmark that creates the most daylight between them.
Which is better for coding, GPT-5.2 or GPT-5.4?
GPT-5.2 has the edge for coding in this comparison, averaging 70.6 versus 57.7. Inside this category, Vibe Code Bench is the benchmark that creates the most daylight between them.
Which is better for math, GPT-5.2 or GPT-5.4?
GPT-5.4 has the edge for math in this comparison, averaging 42.5 versus 35.2. Inside this category, FrontierMath v2 (Tier 4) is the benchmark that creates the most daylight between them.
Which is better for agentic tasks, GPT-5.2 or GPT-5.4?
GPT-5.4 has the edge for agentic tasks in this comparison, averaging 77.2 versus 55.7. Inside this category, OSWorld-Verified is the benchmark that creates the most daylight between them.
Which is better for multimodal and grounded tasks, GPT-5.2 or GPT-5.4?
GPT-5.2 has the edge for multimodal and grounded tasks in this comparison, averaging 80.4 versus 73.2. Inside this category, Design Arena Website is the benchmark that creates the most daylight between them.
Related Comparisons
Explore More
Choose a model with this week’s evidence
Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.
One email each week. Unsubscribe anytime.