Model comparison
GPT-5.4 nano vs GPT-5.6 Luna
Head-to-head evidence from 22 shared benchmark results across 6 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Public leaderboard positions: GPT-5.4 nano #25 (Supported); GPT-5.6 Luna #22 (Estimated). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. GPT-5.4 nano and GPT-5.6 Luna share 22 comparable benchmark results. 4 of 8 categories are comparable. 7 results are unique to GPT-5.4 nano; 19 to GPT-5.6 Luna.
Updated July 23, 2026- Shared results
- 22
- GPT-5.4 nano only
- 7
- GPT-5.6 Luna only
- 19
- Comparable categories
- 4 / 8
Pick GPT-5.6 Luna if you want the stronger benchmark profile. GPT-5.4 nano only becomes the better choice if you want the cheaper token bill.
Confidence note. This is a partial-evidence comparison with 22 shared benchmark results across 6 evidence categories; 4 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
GPT-5.6 Luna has the cleaner BenchAlign overall profile here, landing at 67.17 versus 66.79. It is a real lead, but still close enough that category-level strengths matter more than the headline number.
GPT-5.6 Luna's sharpest advantage is in mathematics, where it averages 73.6 against 21. The single biggest benchmark swing on the page is FrontierMath v2 (Tiers 1-3), 25.860% to 78.600%.
GPT-5.6 Luna is also the more expensive model on tokens at $1.00 input / $6.00 output per 1M tokens, versus $0.20 input / $1.25 output per 1M tokens for GPT-5.4 nano. That is roughly 4.8x on output cost alone. GPT-5.6 Luna gives you the larger context window at 1M, compared with 400K for GPT-5.4 nano.
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-5.4 nano | Δ | GPT-5.6 Luna |
|---|---|---|---|
| Math | GPT-5.4 nano21.0 | Margin→ 52.6 | GPT-5.6 Luna73.6 |
| Knowledge | GPT-5.4 nano43.8 | Margin→ 48.5 | GPT-5.6 Luna92.3 |
| Agentic | GPT-5.4 nano42.9 | Margin→ 41.2 | GPT-5.6 Luna84.1 |
| Multimodal | GPT-5.4 nano66.1 | Margin→ 12.3 | GPT-5.6 Luna78.4 |
| Coding | GPT-5.4 nanoNot measured | MarginNo overlap | GPT-5.6 Luna62.7 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
FrontierMath v2 (Tiers 1-3)
MathA 25.860%B 78.600%Winner: GPT-5.6 LunaΔ 52.7FrontierMath v2 (Tiers 1-3): GPT-5.4 nano scored 25.860%; GPT-5.6 Luna scored 78.600%. GPT-5.6 Luna wins this benchmark. - Source ↗
FrontierMath v2 (Tier 4)
MathA 6.250%B 58.500%Winner: GPT-5.6 LunaΔ 52.3FrontierMath v2 (Tier 4): GPT-5.4 nano scored 6.250%; GPT-5.6 Luna scored 58.500%. GPT-5.6 Luna wins this benchmark. - Source ↗
Terminal-Bench 2.0
AgenticA 46.3%B 84.7%Winner: GPT-5.6 LunaΔ 38.4Terminal-Bench 2.0: GPT-5.4 nano scored 46.3%; GPT-5.6 Luna scored 84.7%. GPT-5.6 Luna wins this benchmark. - Source ↗
MMMU-Pro
MultimodalA 66.1%B 78.4%Winner: GPT-5.6 LunaΔ 12.3MMMU-Pro: GPT-5.4 nano scored 66.1%; GPT-5.6 Luna scored 78.4%. GPT-5.6 Luna wins this benchmark. - Source ↗
GPQA
KnowledgeA 82.8%B 92.3%Winner: GPT-5.6 LunaΔ 9.5GPQA: GPT-5.4 nano scored 82.8%; GPT-5.6 Luna scored 92.3%. GPT-5.6 Luna wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | GPT-5.4 nano | GPT-5.6 Luna | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | GPT-5.4 nano$0.2 input / $1.25 output | GPT-5.6 Luna$1 input / $6 output | GPT-5.4 nano has the lower combined listed price. |
| Generation speedtokens per second | GPT-5.4 nano191 tok/s | GPT-5.6 LunaNot available | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | GPT-5.4 nano3.64 s | GPT-5.6 LunaNot available | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | GPT-5.4 nano400K | GPT-5.6 Luna1M | GPT-5.6 Luna lists the larger context window. |
Benchmark Deep Dive
AgenticGPT-5.6 Luna wins18 benchmarks
| Benchmark | GPT-5.4 nano | GPT-5.6 Luna | Result |
|---|---|---|---|
| Terminal-Bench 2.0Source | 46.3% | 84.7% | GPT-5.6 Luna leads |
| OSWorld-VerifiedSource | 39% | — | Not comparable |
| MCP AtlasSource | 56.1% | — | Not comparable |
| ToolathlonSource | 35.5% | 53.4% | GPT-5.6 Luna leads |
| τ²-bench resultsSource | 76% | — | Not comparable |
| AA Agentic IndexSource | 27.5% | 45.6% | GPT-5.6 Luna leads |
| APEX-Agents-AASource | 24.9% | 35.8% | GPT-5.6 Luna leads |
| GDPval-AASource | 30.0% | 54.2% | GPT-5.6 Luna leads |
| GDPval-AASource | 1100 | 1584 | GPT-5.6 Luna leads |
| BrowseCompSource | — | 83.3% | Not comparable |
| OSWorld 2.0Source | — | 45.6% | Not comparable |
| CyberGymSource | — | 77.9% | Not comparable |
| ExploitGymSource | — | 12.4% | Not comparable |
| AA Harvey LABSource | — | 87.9% | Not comparable |
| AA ITBenchSource | — | 40.3% | Not comparable |
| AA Tau3 BankingSource | — | 27.2% | Not comparable |
| AA AutomationBenchSource | — | 42.2% | Not comparable |
| aaTerminalBench21Source | — | 80.9% | Not comparable |
Coding8 benchmarks
| Benchmark | GPT-5.4 nano | GPT-5.6 Luna | Result |
|---|---|---|---|
| Vibe Code BenchSource | 26.10% | — | Not comparable |
| AA Coding IndexSource | 56.1% | 71.5% | GPT-5.6 Luna leads |
| AA-SciCodeSource | 46.9% | 52.5% | GPT-5.6 Luna leads |
| SWE-bench ProSource | — | 62.7% | Not comparable |
| Terminal-Bench 2.0Source | — | 84.7% | Not comparable |
| deepSweSource | — | 67.2% | Not comparable |
| FrontierCode 1.1 ExtendedSource | — | 55.1% | Not comparable |
| cursorBench32Source | — | 61.1% | Not comparable |
Reasoning3 benchmarks
KnowledgeGPT-5.6 Luna wins12 benchmarks
| Benchmark | GPT-5.4 nano | GPT-5.6 Luna | Result |
|---|---|---|---|
| GPQASource | 82.8% | 92.3% | GPT-5.6 Luna leads |
| HLESource | 37.7% | — | Not comparable |
| HLE w/o toolsSource | 24.3% | — | Not comparable |
| Artificial Analysis Intelligence IndexSource | 38.2% | 51.2% | GPT-5.6 Luna leads |
| AA-GPQA DiamondSource | 81.7% | 91.1% | GPT-5.6 Luna leads |
| AA-HLESource | 26.5% | 37.2% | GPT-5.6 Luna leads |
| AA-Omniscience IndexSource | -29.5% | -11.2% | GPT-5.6 Luna leads |
| AA-Omniscience AccuracySource | 25.4% | 41.5% | GPT-5.6 Luna leads |
| AA-Omniscience Hallucination RateSource | 73.6% | 90.1% | GPT-5.4 nano leads |
| GPQA-DSource | — | 92.3% | Not comparable |
| HealthBench ProfessionalSource | — | 55.7% | Not comparable |
| HealthBench HardSource | — | 32.0% | Not comparable |
MathGPT-5.6 Luna wins3 benchmarks
MultimodalGPT-5.6 Luna wins3 benchmarks
Inst. Following1 benchmarks
| Benchmark | GPT-5.4 nano | GPT-5.6 Luna | Result |
|---|---|---|---|
| AA-IFBenchSource | 75.9% | — | Not comparable |
Frequently Asked Questions (5)
Which is better, GPT-5.4 nano or GPT-5.6 Luna?
GPT-5.6 Luna is ahead on BenchLM's BenchAlign leaderboard, 67.17 to 66.79. The biggest single separator in this matchup is FrontierMath v2 (Tiers 1-3), where the scores are 25.860% and 78.600%.
Which is better for knowledge tasks, GPT-5.4 nano or GPT-5.6 Luna?
GPT-5.6 Luna has the edge for knowledge tasks in this comparison, averaging 92.3 versus 43.8. Inside this category, AA-Omniscience Index is the benchmark that creates the most daylight between them.
Which is better for math, GPT-5.4 nano or GPT-5.6 Luna?
GPT-5.6 Luna has the edge for math in this comparison, averaging 73.6 versus 21. Inside this category, FrontierMath v2 (Tiers 1-3) is the benchmark that creates the most daylight between them.
Which is better for agentic tasks, GPT-5.4 nano or GPT-5.6 Luna?
GPT-5.6 Luna has the edge for agentic tasks in this comparison, averaging 84.1 versus 42.9. Inside this category, GDPval-AA is the benchmark that creates the most daylight between them.
Which is better for multimodal and grounded tasks, GPT-5.4 nano or GPT-5.6 Luna?
GPT-5.6 Luna has the edge for multimodal and grounded tasks in this comparison, averaging 78.4 versus 66.1. Inside this category, AA-MMMU-Pro 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.