GPT-5.5 vs GPT-6 Luna
Full profiles: GPT-5.5 (OpenAI) and GPT-6 Luna (OpenAI): every benchmark, price and rank.
Updated · Data from 15 sources
As of 9 Nov 2026, GPT-5.5 is ahead of GPT-6 Luna on quality (64.3 vs 56.5 on the BenchLeader Index), but GPT-6 Luna is 56× cheaper at $0.200 vs $11.25 per 1M tokens blended. GPT-5.5 leads in every category we measure both in.
Key facts (as of 9 Nov 2026)
- GPT-5.5 scores 64.3 on the BenchLeader Index and GPT-6 Luna scores 56.5 (BenchLeader, 9 Nov 2026).
- GPT-5.5 costs $11.25 per 1M tokens blended (list price, 9 Nov 2026).
- GPT-6 Luna costs $0.200 per 1M tokens blended (list price, 9 Nov 2026).
- Of the 44 benchmarks both models have, GPT-5.5 leads 36 (BenchLeader, 9 Nov 2026).
- On Arena WebDev, GPT-5.5 is rated 1512 and GPT-6 Luna 1577 (LMArena, as published by 9 Nov 2026).
GPT-5.5 benchmarks Which to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| Coding | GPT-5.5 | 61 vs 58 category score; SciCode 55.8% vs 54.6% | Medium |
| Agents & tool use | GPT-5.5 | 62 vs 49 category score; LiveBench Agentic Coding 54.0% vs 51.2% | High |
| Maths | GPT-5.5 | 70 vs 64 category score; FrontierMath Tiers 1–3 85.3% vs 79.0% | Medium |
| Reasoning | GPT-5.5 | 68 vs 56 category score; ARC-AGI-2 85.0% vs 59.3% | High |
| Knowledge | GPT-5.5 | 64 vs 45 category score; SimpleQA Verified 63.0% vs 41.4% | High |
| Long documents | Either | 67 vs 66 category score; AA-LCR v1.1 84.3% vs 83.3% | Either |
| Images and other media | GPT-5.5 | 62 vs 58 category score; MMMU-Pro 79.9% vs 79.7% | Medium |
| Chat quality (human preference) | – | Only GPT-6 Luna has an Arena rating (62 category score) | Not comparable |
| Lowest cost | GPT-6 Luna | $0.200 vs $11.25 per 1M blended; agentic coding session $0.008 vs $0.420 | – |
| Fastest first answer | GPT-5.5 | 44.1 s (GPT-5.5 (xhigh)) vs 85.0 s (GPT-6 Luna (max)) to the first answer token | – |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- GPT-5.5 (xhigh) leads on quality: 64.3 vs 56.5.
- GPT-5.5 (xhigh) is stronger in agents & tools, coding, instruction following, knowledge, long context, maths, multimodal, reasoning.
- GPT-6 Luna (max) is stronger in human preference.
- GPT-6 Luna (max) is 56× cheaper ($0.200 vs $11.25 per 1M blended).
At a glance
| Spec | GPT-5.5 | GPT-6 Luna |
|---|---|---|
| Developer | OpenAI | OpenAI |
| Released | 23 Apr 2026 | 22 Sep 2026 |
| Context window (tokens) | 1,050,000 | 1,050,000 |
| Compared at | GPT-5.5 (xhigh) | GPT-6 Luna (max) |
| Open weights | No | No |
| Price in / out / cached input per 1M | $5.00 / $30.00 / $0.500 | $0.100 / $0.500 / $0.010 |
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | GPT-5.5 (xhigh) | GPT-6 Luna (max) | Leader |
|---|---|---|---|
| Reasoning | 68.3 | 55.9 | GPT-5.5 |
| Coding | 61.0 | 57.5 | GPT-5.5 |
| Agents & tools | 61.8 | 48.5 | GPT-5.5 |
| Maths | 69.8 | 64.1 | GPT-5.5 |
| Knowledge | 64.0 | 45.1 | GPT-5.5 |
| Instruction following | 74.0 | – | not comparable |
| Human preference | – | 62.2 | not comparable |
| Multimodal | 62.3 | 57.6 | GPT-5.5 |
| Long context | 67.0 | 66.4 | level |
On one chart
Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.
Not drawn, measured for fewer than two of these models: Instruction following, Human preference.
ModelIndex
Where sources disagree
Across the 44 benchmarks both have, GPT-5.5 leads 36, GPT-6 Luna leads 7 and 1 is a tie. Every overall index that rates both puts GPT-5.5 ahead.
| Overall index | GPT-5.5 (xhigh) | GPT-6 Luna (max) | Leader |
|---|---|---|---|
| BenchLeader Index | 64.3 | 56.5 | GPT-5.5 |
| Arena Text | 1481† | 1441 | GPT-5.5 |
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.
| Measure | GPT-5.5 (xhigh) | GPT-6 Luna (max) |
|---|---|---|
| Blended price per 1M tokens | $11.25 | $0.200 |
| Input / output per 1M | $5.00 / $30.00 | $0.100 / $0.500 |
| Cached input per 1M | $0.500 | $0.010 |
| Price per task: chat reply | $0.011 | $0.0002 |
| Price per task: summarize a 30-page report | $0.078 | $0.0015 |
| Price per task: code edit | $0.075 | $0.0014 |
| Price per task: agentic coding session | $0.420 | $0.0080 |
| Price per task: structured extraction | $0.016 | $0.0003 |
| Agentic coding session, input mostly cached | $0.217 | $0.0040 |
| Cost per benchmark task (AA Index) | – | $0.152 |
| Output tokens per task (AA Index) | – | 75k |
| Of which reasoning | – | 58k |
| Time per task (AA Index) | – | 12 min |
| Output speed | 80 tok/s | 104 tok/s |
| Time to first answer | 44.1 s | 85.0 s |
| Time to a complete answer | 98.5 s | 177.1 s |
Each model is compared at its best-ranked reasoning-effort setting. GPT-5.5 (xhigh) scores 64.3 and takes 44 s to its first answer token; GPT-5.5 (high) scores 63.7 in 23 s, GPT-5.5 scores 61.7 in 2 s. GPT-6 Luna (max) scores 56.5 and takes 85 s to its first answer token; GPT-6 Luna (xhigh) scores 52.1 in 22 s, GPT-6 Luna (high) scores 51.2 in 11 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
174 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | GPT-5.5 (xhigh) | GPT-6 Luna (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 64.3 | 56.5 |
| Agents & tools score | BenchLeader | 61.8 | 48.5 |
| Coding score | BenchLeader | 61.0 | 57.5 |
| Instruction following score | BenchLeader | 74.0 | – |
| Knowledge score | BenchLeader | 64.0 | 45.1 |
| Long context score | BenchLeader | 67.0 | 66.4 |
| Maths score | BenchLeader | 69.8 | 64.1 |
| Multimodal score | BenchLeader | 62.3 | 57.6 |
| Reasoning score | BenchLeader | 68.3 | 55.9 |
| Human preference score | BenchLeader | – | 62.2 |
| Blended price $/M | BenchLeader | $11.25 | $0.200 |
| Output speed | BenchLeader | 80 tok/s | 104 tok/s |
| Time to first answer | BenchLeader | 44.1 s | 85.0 s |
| Context window | BenchLeader | 1,050,000 tokens | 1,050,000 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 90.7%† | 90.5% |
| SimpleBench | SimpleBench | 69.0%† | – |
| Arena Hard Prompts | Arena (LMArena) | 1499† | 1463 |
| LiveBench 2.0 Reasoning | LiveBench | – | 81.8%† |
| LiveBench 1.0 Reasoning | LiveBench | 89.7% | 81.8% |
| GPQA Diamond (AA) | Artificial Analysis | 93.5% | – |
| Humanity's Last Exam (AA) | Artificial Analysis | 45.8% | 38.5% |
| HLE-Diamond (AA) | Artificial Analysis | – | 22.2% |
| GPQA Diamond (Vals) | Vals AI | 93.2% | – |
| Kagi LLM Benchmark | Kagi LLM Benchmark | 88.8%† | – |
| ARC-AGI-1 | ARC Prize | 95.0% | 86.7% |
| ARC-AGI-2 | ARC Prize | 85.0% | 59.3% |
| ARC-AGI-3 | ARC Prize | 0.4%† | 0.6% |
| CritPt | Artificial Analysis | 27.1% | 19.4% |
| CritPt v2 hard subset (AA) | Artificial Analysis | – | 42.5% |
| MysteryMechanism | Vals AI | – | 19.4% |
| Chess Puzzles | Epoch AI Benchmarking Hub | 63.0% | 36.0% |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 56.0% | 7.0% |
| CL-bench Life | Epoch AI Benchmarking Hub | 22.2%† | – |
| LMCA | Epoch AI Benchmarking Hub | 54.3% | 44.5% |
| DTBench | Epoch AI Benchmarking Hub | 96.0% | 90.1% |
| ForecastBench | ForecastBench | 64.5%† | – |
| BTF-3 | Epoch AI Benchmarking Hub | 14.3%† | – |
| Coding | |||
| SciCode | SciCode | 55.8% | 54.6% |
| WeirdML | WeirdML | 84.9% | – |
Show all 174 benchmarks
| Metric | Publisher | GPT-5.5 (xhigh) | GPT-6 Luna (max) |
|---|---|---|---|
| Coding | |||
| FrontierCode | Cognition | 43.0%† | 42.4% |
| GSO-Bench | GSO-Bench | 40.2% | – |
| Arena Coding | Arena (LMArena) | 1510† | 1502 |
| Arena WebDev | Arena (LMArena) | 1512 | 1577 |
| LiveBench 2.0 Coding | LiveBench | – | 70.8%† |
| LiveBench 1.0 Coding | LiveBench | 82.2% | 79.0% |
| LiveCodeBench | Vals AI | 85.3% | – |
| IOI | Vals AI | – | 55.6% |
| SWE-bench (Vals) | Vals AI | 82.6% | – |
| Code Migration | Vals AI | 45.2% | 42.5% |
| Code Migration v1.1 | Vals AI | – | 46.7% |
| ProgramBench | Vals AI | 0.5% | 0.5% |
| SWE-bench Multimodal | Vals AI | – | 37.5% |
| Vibe Code Bench v1.1 | Vals AI | 69.8% | 81.7% |
| SWE Atlas: Codebase QnA | Scale AI SEAL | 45.4% | – |
| SWE Atlas: Refactoring | Scale AI SEAL | 44.8% | – |
| SWE Atlas: Test Writing | Scale AI SEAL | 42.6% | – |
| MirrorCode | Epoch AI Benchmarking Hub | 10.0%† | 26.7% |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | 88.1%† | – |
| DeepSWE v1.1 | Epoch AI Benchmarking Hub | 67.0% | – |
| ALE-Bench | Epoch AI Benchmarking Hub | 1943.0 | 1576.9† |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 14.7% | 12.6% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 84.3% | – |
| Startrise Frontend | Startrise | – | 69.3%† |
| GameBench 2 Overall | DuelLab GameBench 2 | 49.0† | 47.1† |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | 59.7 | 55.8† |
| Design Arena Website | Design Arena | 1253† | 1265† |
| FrontierCode 1.1 (AA) | Artificial Analysis | – | 41.9% |
| FrontierCode 2.0 (AA) | Artificial Analysis | – | 25.2% |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 84.7%† | 15.2% |
| OSWorld-Verified 2.0 | OSWorld | 13.0% | – |
| Remote Labor Index | Scale AI / CAIS | 6.3%† | – |
| APEX-Agents | Mercor | 55.1%† | 44.3% |
| Arena Agent | Arena (LMArena) | – | 1.5 |
| LiveBench Long Agentic | LiveBench | – | 24.7%† |
| LiveBench Browser | LiveBench | – | 70.3%† |
| LiveBench Agentic Coding | LiveBench | 54.0% | 51.2% |
| Terminal-Bench Hard | Artificial Analysis | 60.6% | – |
| τ²-Bench Telecom (AA) | Artificial Analysis | 93.9% | – |
| Terminal-Bench 2.1 (Vals) | Vals AI | 76.4%† | 73.0% |
| MCP Atlas | Scale AI SEAL | 75.3% | – |
| HiL-Bench | Scale AI SEAL | 39.7%† | – |
| τ³-bench Banking | τ²-bench | 44.6% | – |
| GDPval-AA v2.1, normalised score | Artificial Analysis | – | 46.6% |
| GDPval-AA v2.1 | Artificial Analysis | – | 1444 |
| τ³-Banking (AA) | Artificial Analysis | 39.0% | – |
| ITBench SRE (AA) | Artificial Analysis | 45.8% | – |
| Analyst Agent (AA) | Artificial Analysis | 50.0% | 43.8% |
| APEX-Agents (AA) | Artificial Analysis | 37.7% | – |
| CyberBench | Vals AI | – | 76.3% |
| Finance Agent v2 | Vals AI | 51.8% | 49.9% |
| Finance Agent v2.1 | Vals AI | – | 56.8% |
| Harvey's Legal Agent Benchmark v1.1 | Vals AI | – | 4.2% |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 3.8% | 2.9% |
| Legal Research Bench | Vals AI | 40.4% | 30.3% |
| SkillsBench | Vals AI | 62.2% | – |
| SREBench | Vals AI | 3.8% | 2.7% |
| Tax Agent Bench | Vals AI | 21.8% | 19.9% |
| Terminal-Bench 2.0 (Vals) | Vals AI | 73.2%† | – |
| Terminal-Bench 4.0 (Vals) | Vals AI | – | 13.6% |
| Terminal-Bench Science | Vals AI | – | 4.3% |
| Time Horizon Index: KSP | Vals AI | 8.3% | – |
| Arena Search | Arena (LMArena) | 1226† | 1211 |
| DeepResearch Bench | DeepResearch Bench | 54.0%† | – |
| PostTrainBench | Epoch AI Benchmarking Hub | 27.2% | – |
| ExploitBench | Epoch AI Benchmarking Hub | 47.4%† | – |
| GBAEval | Epoch AI Benchmarking Hub | 53.2%† | – |
| Vending-Bench 2 | Andon Labs | 7523.8† | – |
| AutomationBench | Artificial Analysis | 47.3% | 53.2% |
| GDP.pdf (AA) | Artificial Analysis | 21.2% | 22.8% |
| Harvey LAB-AA v1.1 | Harvey | – | 3.3% |
| EnterpriseOps-Gym | Artificial Analysis | 46.6% | – |
| AA-Briefcase v1.1 | Artificial Analysis | 1135 | 1337 |
| GDPx | GDPx by SemiAnalysis | – | 5.1% |
| Design Arena Agentic | Design Arena | 1108† | 1149† |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | – | 8.6% |
| CWE-Bench (AA) | Artificial Analysis | – | 56.7% |
| DeepSec-Bench (AA) | Artificial Analysis | – | 23.5% |
| CyberGym end-to-end (AA) | Artificial Analysis | – | 77.9% |
| FACTS Search | – | 86.8%† | |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 85.3% | 79.0% |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 72.5% | 56.1% |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 84.4%† | 98.9% |
| ProofBench | Vals AI | 50.0% | 64.0%† |
| LiveBench Mathematics | LiveBench | 95.9% | 89.1% |
| AIME 2026 | MathArena | 100.0% | – |
| HMMT February 2026 | MathArena | 98.5% | – |
| USAMO 2026 | MathArena | 98.2% | – |
| MathArena Apex | MathArena | 80.2% | – |
| Arena Maths | Arena (LMArena) | 1500† | 1450 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 63.0% | 41.4% |
| LiveBench 2.0 Data Analysis | LiveBench | – | 73.4%† |
| LiveBench Research | LiveBench | – | 48.5%† |
| LiveBench 1.0 Data Analysis | LiveBench | 81.6% | 73.4% |
| AA-Omniscience v1.1 | Artificial Analysis | – | -8.3 |
| MMLU-Pro | Vals AI | 88.1% | – |
| LegalBench | Vals AI | 86.5% | – |
| CorpFin | Vals AI | 68.4% | – |
| TaxEval | Vals AI | 75.0% | – |
| PRBench Legal | Scale AI SEAL | – | 42.9%† |
| BioMysteryBench | Vals AI | – | 61.5% |
| CaseLaw v2 | Vals AI | 66.2% | – |
| ClaimBench | Vals AI | – | 99.2% |
| Excel Modeling Benchmark | Vals AI | 64.5% | 68.5% |
| Excel Modeling Benchmark v1.1 | Vals AI | – | 70.0% |
| MedCode | Vals AI | 49.1% | 44.7% |
| MedScribe | Vals AI | 86.9% | 83.7% |
| Public Benefits Bench v1.1 | Vals AI | 60.9% | 57.6% |
| DrugDiscoveryBench | Scale AI SEAL | 51.6% | – |
| EBR-bench | Epoch AI Benchmarking Hub | 34.3% | – |
| GDP.pdf | Epoch AI Benchmarking Hub | 26.0% | 23.0% |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | – | 39.6% |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | – | 20.6% |
| HealthBench Professional (AA) | OpenAI | – | 61.9% |
| Instruction following | |||
| LiveBench Language | LiveBench | 87.4% | 73.8% |
| LiveBench Instruction Following | LiveBench | 70.7% | 55.9% |
| IFBench | Artificial Analysis | 75.8% | – |
| Arena Instruction Following | Arena (LMArena) | 1475† | 1436 |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1476† | 1441 |
| EQ-Bench 4 | EQ-Bench | 1295† | – |
| Arena Creative Writing | Arena (LMArena) | 1448† | 1407 |
| Arena Multi-turn | Arena (LMArena) | 1484† | 1442 |
| Arena Longer Queries | Arena (LMArena) | 1487† | 1448 |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1296† | 1211 |
| LiveBench Design | LiveBench | – | 76.1%† |
| MMMU-Pro | Artificial Analysis | 79.9% | 79.7% |
| MMMU-Pro (Vals) | Vals AI | 88.3% | – |
| MortgageTax | Vals AI | 68.8% | – |
| SAGE | Vals AI | – | 83.4% |
| Blueprint-Bench 2 | Epoch AI Benchmarking Hub | 36.2%† | 31.2%† |
| Code2Video-AA | Artificial Analysis | 1000 | 1030 |
| AA-3D | Artificial Analysis | – | 1000 |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 84.3% | 83.3% |
| Arena Document | Arena (LMArena) | 1479† | – |
| MLCR | Artificial Analysis | 28.9% | 26.1% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 160.0† | 156.7† |
| LiveBench 2.0 | LiveBench | – | 60.4%† |
| LiveBench 1.0 | LiveBench | 80.2% | 72.0% |
| AA Intelligence Index v5.0 | Artificial Analysis | – | 27.9 |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 38.4 | 38.1 |
| AA Progress Index v1 | Artificial Analysis | 320 | 317.1 |
| Vals Index | Vals AI | – | 52.3 |
| Vals Finance Index | Vals AI | – | 53.0% |
| Vals RSI Index | Vals AI | 16.5% | – |
| Vals Science Index | Vals AI | – | 28.4% |
| Vals Multimodal Index | Vals AI | 68.1% | – |
| Vals Router Index | Vals AI | – | 52.3%† |
| AA Cyber Index | Artificial Analysis | – | 52.7 |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | 16.3% | 10.8%† |
Data as of 2026-11-09. Best configuration of each model; every score links to its source on the model pages.
DuelLab GameBench 2, release gb2-20261103-eaea31df94e6 (https://benchmarks.duellab.org/releases/gb2-20261103-eaea31df94e6/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
What changed
- GPT-6 Luna (max): 1 score revision.
- GPT-6 Luna (max): 1 score revision.
- GPT-5.5 (xhigh): 1 score revision.
- GPT-6 Luna (max): 1 score revision.
- GPT-6 Luna (max): 1 score revision.
- GPT-6 Luna (max): 1 new result (MirrorCode); 2 score revisions.
- GPT-6 Luna (max): BenchLeader Index 58.6 → 56.6.
- GPT-5.5 (xhigh): 1 new result (AA Progress Index v1).
- GPT-6 Luna (max): 5 new results (AA Intelligence Index v5.0, AA Progress Index v1, HLE-Diamond (AA)…); 4 score revisions; cost per task moved from $0.068/M to $0.15/M (124% up).
- GPT-5.5 (xhigh): cost per task moved from $2.63/M to $2.10/M (20% down).
Related comparisons
- GPT-5.5 vs Claude Fable 5.2
- GPT-5.5 vs Claude Opus 5.5
- GPT-5.5 vs Gemini 4 Carbon
- GPT-6 Luna vs GPT-6.1 Sol
- GPT-6 Luna vs Gemini 4 Argon
- GPT-6 Luna vs Claude Sonnet 5.5
- GPT-5.5 vs GPT-6.1 Sol
- GPT-5.5 vs Claude Fable 5.1
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
GPT-5.5 vs GPT-6 Luna: questions
- Is GPT-5.5 better than GPT-6 Luna?
- GPT-5.5 (xhigh) leads on quality: 64.3 vs 56.5. The BenchLeader Index combines every independent quality benchmark; GPT-5.5 (xhigh) is ahead overall as of 9 Nov 2026, but check the category scores for your use.
- GPT-5.5 vs GPT-6 Luna: which is cheaper, and will that change?
- GPT-6 Luna is cheaper today: $0.200 against $11.25 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is GPT-5.5 or GPT-6 Luna better for coding?
- GPT-5.5 scores higher in coding (61 vs 58 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1512 for GPT-5.5 against 1577 for GPT-6 Luna.
- Which is faster, GPT-5.5 or GPT-6 Luna?
- GPT-6 Luna streams faster (104 vs 80 output tokens per second, Artificial Analysis 7-day median).
- Is GPT-5.5 or GPT-6 Luna better for agents and tool use?
- GPT-5.5 scores higher in agentic tasks (62 vs 49 on the category index, where 50 is average).
More questions (4)
- Which is better for long documents?
- They are level on long-context benchmarks (67 vs 66). Both accept 1,050,000 tokens of context.
- Which has lower latency?
- GPT-5.5 answers first: 44.1 s against 85.0 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
- Do independent benchmarks agree on GPT-5.5 vs GPT-6 Luna?
- Across the 44 benchmarks both have, GPT-5.5 leads 36, GPT-6 Luna leads 7 and 1 is a tie. Every overall index that rates both puts GPT-5.5 ahead.
- When was GPT-6 Luna released?
- GPT-6 Luna was released on 22 Sep 2026.