Model comparison
GPT-5.6 Luna vs GPT-5.6 Terra
Head-to-head evidence from 41 shared benchmark results across 6 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Sibling matchup inside the GPT-5.6 family.
Public leaderboard positions: GPT-5.6 Luna #22 (Estimated); GPT-5.6 Terra #11 (Estimated). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. GPT-5.6 Luna and GPT-5.6 Terra share 41 comparable benchmark results. 5 of 8 categories are comparable. 0 results are unique to GPT-5.6 Luna; 3 to GPT-5.6 Terra.
Updated July 23, 2026- Shared results
- 41
- GPT-5.6 Luna only
- 0
- GPT-5.6 Terra only
- 3
- Comparable categories
- 5 / 8
GPT-5.6 Luna makes more sense if you want the cheaper token bill, while GPT-5.6 Terra is the cleaner fit if mathematics is the priority.
Confidence note. This is a partial-evidence comparison with 41 shared benchmark results across 6 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-5.6 Luna and GPT-5.6 Terra sit in the same GPT-5.6 family. This page is less about two unrelated model lineages and more about how the siblings trade off on benchmark shape, token costs, and practical limits like context window.
GPT-5.6 Terra is clearly ahead on the BenchAlign aggregate, 72.57 to 67.17. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
GPT-5.6 Terra's sharpest advantage is in mathematics, where it averages 80.8 against 73.6. The single biggest benchmark swing on the page is FrontierMath v2 (Tier 4), 58.500% to 68.300%.
GPT-5.6 Terra is also the more expensive model on tokens at $2.50 input / $15.00 output per 1M tokens, versus $1.00 input / $6.00 output per 1M tokens for GPT-5.6 Luna. That is roughly 2.5x on output cost alone.
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-5.6 Luna | Δ | GPT-5.6 Terra |
|---|---|---|---|
| Math | GPT-5.6 Luna73.6 | Margin→ 7.2 | GPT-5.6 Terra80.8 |
| Agentic | GPT-5.6 Luna84.1 | Margin→ 3.3 | GPT-5.6 Terra87.4 |
| Multimodal | GPT-5.6 Luna78.4 | Margin→ 2.3 | GPT-5.6 Terra80.7 |
| Coding | GPT-5.6 Luna62.7 | Margin→ 0.7 | GPT-5.6 Terra63.4 |
| Knowledge | GPT-5.6 Luna92.3 | Margin→ 0.6 | GPT-5.6 Terra92.9 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
FrontierMath v2 (Tier 4)
MathA 58.500%B 68.300%Winner: GPT-5.6 TerraΔ 9.8FrontierMath v2 (Tier 4): GPT-5.6 Luna scored 58.500%; GPT-5.6 Terra scored 68.300%. GPT-5.6 Terra wins this benchmark. - Source ↗
FrontierMath v2 (Tiers 1-3)
MathA 78.600%B 84.900%Winner: GPT-5.6 TerraΔ 6.3FrontierMath v2 (Tiers 1-3): GPT-5.6 Luna scored 78.600%; GPT-5.6 Terra scored 84.900%. GPT-5.6 Terra wins this benchmark. - Source ↗
BrowseComp
AgenticA 83.3%B 87.5%Winner: GPT-5.6 TerraΔ 4.2BrowseComp: GPT-5.6 Luna scored 83.3%; GPT-5.6 Terra scored 87.5%. GPT-5.6 Terra wins this benchmark. - Source ↗
Terminal-Bench 2.0
AgenticA 84.7%B 87.4%Winner: GPT-5.6 TerraΔ 2.7Terminal-Bench 2.0: GPT-5.6 Luna scored 84.7%; GPT-5.6 Terra scored 87.4%. GPT-5.6 Terra wins this benchmark. - Source ↗
MMMU-Pro
MultimodalA 78.4%B 80.7%Winner: GPT-5.6 TerraΔ 2.3MMMU-Pro: GPT-5.6 Luna scored 78.4%; GPT-5.6 Terra scored 80.7%. GPT-5.6 Terra wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | GPT-5.6 Luna | GPT-5.6 Terra | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | GPT-5.6 Luna$1 input / $6 output | GPT-5.6 Terra$2.5 input / $15 output | GPT-5.6 Luna has the lower combined listed price. |
| Generation speedtokens per second | GPT-5.6 LunaNot available | GPT-5.6 TerraNot available | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | GPT-5.6 LunaNot available | GPT-5.6 TerraNot available | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | GPT-5.6 Luna1M | GPT-5.6 Terra1M | Listed context windows are equal. |
Benchmark Deep Dive
AgenticGPT-5.6 Terra wins17 benchmarks
| Benchmark | GPT-5.6 Luna | GPT-5.6 Terra | Result |
|---|---|---|---|
| Terminal-Bench 2.0Source | 84.7% | 87.4% | GPT-5.6 Terra leads |
| BrowseCompSource | 83.3% | 87.5% | GPT-5.6 Terra leads |
| OSWorld 2.0Source | 45.6% | 50.2% | GPT-5.6 Terra leads |
| CyberGymSource | 77.9% | 81.8% | GPT-5.6 Terra leads |
| ExploitGymSource | 12.4% | 23.2% | GPT-5.6 Terra leads |
| ToolathlonSource | 53.4% | 53.1% | GPT-5.6 Luna leads |
| AA Agentic IndexSource | 45.6% | 47.4% | GPT-5.6 Terra leads |
| GDPval-AASource | 54.2% | 54.1% | GPT-5.6 Luna leads |
| GDPval-AASource | 1584 | 1581 | GPT-5.6 Luna leads |
| AA Harvey LABSource | 87.9% | 85.2% | GPT-5.6 Luna leads |
| AA ITBenchSource | 40.3% | 51.0% | GPT-5.6 Terra leads |
| AA Tau3 BankingSource | 27.2% | 31.8% | GPT-5.6 Terra leads |
| AA AutomationBenchSource | 42.2% | 45.6% | GPT-5.6 Terra leads |
| aaTerminalBench21Source | 80.9% | 88% | GPT-5.6 Terra leads |
| APEX-Agents-AASource | 35.8% | 38.9% | GPT-5.6 Terra leads |
| τ²-bench resultsSource | — | 86.3% | Not comparable |
| terminalBenchHardSource | — | 57.6% | Not comparable |
CodingGPT-5.6 Terra wins7 benchmarks
| Benchmark | GPT-5.6 Luna | GPT-5.6 Terra | Result |
|---|---|---|---|
| SWE-bench ProSource | 62.7% | 63.4% | GPT-5.6 Terra leads |
| Terminal-Bench 2.0Source | 84.7% | 87.4% | GPT-5.6 Terra leads |
| deepSweSource | 67.2% | 69.6% | GPT-5.6 Terra leads |
| FrontierCode 1.1 ExtendedSource | 55.1% | 55.8% | GPT-5.6 Terra leads |
| cursorBench32Source | 61.1% | 64.9% | GPT-5.6 Terra leads |
| AA Coding IndexSource | 71.5% | 76.7% | GPT-5.6 Terra leads |
| AA-SciCodeSource | 52.5% | 53.9% | GPT-5.6 Terra leads |
Reasoning3 benchmarks
KnowledgeGPT-5.6 Terra wins10 benchmarks
| Benchmark | GPT-5.6 Luna | GPT-5.6 Terra | Result |
|---|---|---|---|
| GPQASource | 92.3% | 92.9% | GPT-5.6 Terra leads |
| GPQA-DSource | 92.3% | 92.9% | GPT-5.6 Terra leads |
| HealthBench ProfessionalSource | 55.7% | 57.7% | GPT-5.6 Terra leads |
| HealthBench HardSource | 32.0% | 32.7% | GPT-5.6 Terra leads |
| Artificial Analysis Intelligence IndexSource | 51.2% | 55.0% | GPT-5.6 Terra leads |
| AA-GPQA DiamondSource | 91.1% | 92.5% | GPT-5.6 Terra leads |
| AA-HLESource | 37.2% | 41.8% | GPT-5.6 Terra leads |
| AA-Omniscience IndexSource | -11.2% | -0.2% | GPT-5.6 Terra leads |
| AA-Omniscience AccuracySource | 41.5% | 45.9% | GPT-5.6 Terra leads |
| AA-Omniscience Hallucination RateSource | 90.1% | 85.2% | GPT-5.6 Terra leads |
MathGPT-5.6 Terra wins3 benchmarks
MultimodalGPT-5.6 Terra wins3 benchmarks
Inst. Following1 benchmarks
| Benchmark | GPT-5.6 Luna | GPT-5.6 Terra | Result |
|---|---|---|---|
| AA-IFBenchSource | — | 71.2% | Not comparable |
Frequently Asked Questions (6)
Which is better, GPT-5.6 Luna or GPT-5.6 Terra?
GPT-5.6 Luna and GPT-5.6 Terra are sibling variants in the GPT-5.6 family, so the right pick depends on whether you value the better benchmark line, cheaper tokens, or the larger context window. GPT-5.6 Terra is ahead on BenchLM's BenchAlign leaderboard 72.57 to 67.17.
Which is better for knowledge tasks, GPT-5.6 Luna or GPT-5.6 Terra?
GPT-5.6 Terra has the edge for knowledge tasks in this comparison, averaging 92.9 versus 92.3. Inside this category, AA-Omniscience Index is the benchmark that creates the most daylight between them.
Which is better for coding, GPT-5.6 Luna or GPT-5.6 Terra?
GPT-5.6 Terra has the edge for coding in this comparison, averaging 63.4 versus 62.7. Inside this category, AA Coding Index is the benchmark that creates the most daylight between them.
Which is better for math, GPT-5.6 Luna or GPT-5.6 Terra?
GPT-5.6 Terra has the edge for math in this comparison, averaging 80.8 versus 73.6. Inside this category, FrontierMath v2 (Tier 4) is the benchmark that creates the most daylight between them.
Which is better for agentic tasks, GPT-5.6 Luna or GPT-5.6 Terra?
GPT-5.6 Terra has the edge for agentic tasks in this comparison, averaging 87.4 versus 84.1. Inside this category, ExploitGym is the benchmark that creates the most daylight between them.
Which is better for multimodal and grounded tasks, GPT-5.6 Luna or GPT-5.6 Terra?
GPT-5.6 Terra has the edge for multimodal and grounded tasks in this comparison, averaging 80.7 versus 78.4. Inside this category, MMMU-Pro w/ Python 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.