GPT-5.6 Terra vs o3
Full profiles: GPT-5.6 Terra (OpenAI) and o3 (OpenAI): every benchmark, price and rank.
Updated · Data from 19 sources
As of 2 Nov 2026, GPT-5.6 Terra and o3 are neck and neck on quality (61.1 vs 61.0 on the BenchLeader Index, inside the ±6.0 margin), and o3 is 1.3× cheaper at $3.50 vs $4.50 per 1M tokens blended. Choose GPT-5.6 Terra for multimodal (63 vs 57), instruction following (70 vs 65), and long context (66 vs 62); choose o3 for fast first answers (6.8 s vs 148.7 s) and agents & tools (70 vs 59), knowledge (56 vs 49), and maths (71 vs 69).
Key facts (as of 2 Nov 2026)
- GPT-5.6 Terra scores 61.1 on the BenchLeader Index and o3 scores 61.0 (BenchLeader, 2 Nov 2026).
- GPT-5.6 Terra costs $4.50 per 1M tokens blended (list price, 2 Nov 2026).
- o3 costs $3.50 per 1M tokens blended (list price, 2 Nov 2026).
- Of the 9 benchmarks both models have, GPT-5.6 Terra leads 8 (BenchLeader, 2 Nov 2026).
GPT-5.6 Terra 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-5.6 Terra | 65 vs 62 category score | Medium |
| Agents & tool use | o3 | 70 vs 59 category score; Terminal-Bench Hard 37.1% vs 57.6% | High |
| Maths | o3 | 71 vs 69 category score | Medium |
| Reasoning | Either | 61 vs 61 category score; GPQA Diamond (AA) 92.5% vs 82.7% | Either |
| Knowledge | o3 | 56 vs 49 category score | High |
| Long documents | GPT-5.6 Terra | 66 vs 62 category score; AA-LCR v1.1 83.0% vs 74.7% | Medium |
| Images and other media | GPT-5.6 Terra | 63 vs 57 category score; MMMU-Pro 80.7% vs 70.1% | Medium |
| Following instructions | GPT-5.6 Terra | 70 vs 65 category score; IFBench 71.2% vs 71.4% | Medium |
| Chat quality (human preference) | – | Only o3 has an Arena rating (61 category score) | Not comparable |
| Lowest cost | o3 | $3.50 vs $4.50 per 1M blended; agentic coding session $0.152 vs $0.168 | – |
| Fastest first answer | o3 | 6.8 s (o3) vs 148.7 s (GPT-5.6 Terra (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-5.6 Terra (max) and o3 are neck and neck on quality (61.1 vs 61.0, within the ±6.0 margin).
- GPT-5.6 Terra (max) is stronger in coding, instruction following, long context, multimodal, reasoning.
- o3 is stronger in agents & tools, knowledge, maths, human preference.
- o3 is 1.3× cheaper ($3.50 vs $4.50 per 1M blended).
At a glance
| Spec | GPT-5.6 Terra | o3 |
|---|---|---|
| Developer | OpenAI | OpenAI |
| Released | 9 Jul 2026 | 16 Apr 2025 |
| Context window (tokens) | 1,050,000 | 200,000 |
| Compared at | GPT-5.6 Terra (max) | o3 |
| Open weights | No | No |
| Price in / out / cached input per 1M | $2.00 / $12.00 / $0.200 | $2.00 / $8.00 / $0.500 |
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-5.6 Terra (max) | o3 | Leader |
|---|---|---|---|
| Reasoning | 61.4 | 60.9 | level |
| Coding | 64.9 | 61.8 | GPT-5.6 Terra |
| Agents & tools | 58.8 | 70.3 | o3 |
| Maths | 68.6 | 70.5 | o3 |
| Knowledge | 48.6 | 56.1 | o3 |
| Instruction following | 69.9 | 65.1 | GPT-5.6 Terra |
| Human preference | – | 61.1 | not comparable |
| Multimodal | 63.3 | 56.7 | GPT-5.6 Terra |
| Long context | 66.4 | 62.1 | GPT-5.6 Terra |
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: Human preference.
ModelIndex
Where sources disagree
Across the 9 benchmarks both have, GPT-5.6 Terra leads 8, o3 leads 1. Every overall index that rates both puts GPT-5.6 Terra ahead.
| Overall index | GPT-5.6 Terra (max) | o3 | Leader |
|---|---|---|---|
| BenchLeader Index | 61.1 | 61.0 | neck and neck |
| Arena Text | 1467† | 1432 | GPT-5.6 Terra |
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-5.6 Terra (max) | o3 |
|---|---|---|
| Blended price per 1M tokens | $4.50 | $3.50 |
| Input / output per 1M | $2.00 / $12.00 | $2.00 / $8.00 |
| Cached input per 1M | $0.200 | $0.500 |
| Price per task: chat reply | $0.0044 | $0.0032 |
| Price per task: summarize a 30-page report | $0.031 | $0.029 |
| Price per task: code edit | $0.030 | $0.024 |
| Price per task: agentic coding session | $0.168 | $0.152 |
| Price per task: structured extraction | $0.0064 | $0.0056 |
| Agentic coding session, input mostly cached | $0.087 | $0.085 |
| Output speed | 130 tok/s | 160 tok/s |
| Time to first answer | 148.7 s | 6.8 s |
| Time to a complete answer | 152.5 s | 9.9 s |
Each model is compared at its best-ranked reasoning-effort setting. GPT-5.6 Terra (max) scores 61.1 and takes 149 s to its first answer token; GPT-5.6 Terra (xhigh) scores 60.3 in 22 s, GPT-5.6 Terra (high) scores 57.0 in 5 s. o3 scores 61.0 and takes 7 s to its first answer token; o3 (high) scores 53.8 in 7 s, o3 (medium) scores 53.3 in 7 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
162 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-5.6 Terra (max) | o3 |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 61.1 | 61.0 |
| Agents & tools score | BenchLeader | 58.8 | 70.3 |
| Coding score | BenchLeader | 64.9 | 61.8 |
| Instruction following score | BenchLeader | 69.9 | 65.1 |
| Knowledge score | BenchLeader | 48.6 | 56.1 |
| Long context score | BenchLeader | 66.4 | 62.1 |
| Maths score | BenchLeader | 68.6 | 70.5 |
| Multimodal score | BenchLeader | 63.3 | 56.7 |
| Reasoning score | BenchLeader | 61.4 | 60.9 |
| Human preference score | BenchLeader | – | 61.1 |
| Blended price $/M | BenchLeader | $4.50 | $3.50 |
| Output speed | BenchLeader | 130 tok/s | 160 tok/s |
| Time to first answer | BenchLeader | 148.7 s | 6.8 s |
| Context window | BenchLeader | 1,050,000 tokens | 200,000 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 93.3% | 81.8%† |
| Humanity's Last Exam | Scale AI / CAIS | – | 20.3%† |
| SimpleBench | SimpleBench | 48.9%† | 53.1%† |
| Arena Hard Prompts | Arena (LMArena) | 1490† | 1441 |
| LiveBench 2.0 Reasoning | LiveBench | 90.6%† | – |
| LiveBench 1.0 Reasoning | LiveBench | 90.6% | – |
| GPQA Diamond (AA) | Artificial Analysis | 92.5% | 82.7% |
| Humanity's Last Exam (AA) | Artificial Analysis | 42.9% | 20.1% |
| HLE-Diamond (AA) | Artificial Analysis | 28.8% | – |
| GPQA Diamond (Vals) | Vals AI | 90.9%† | 84.1%† |
| Kagi LLM Benchmark | Kagi LLM Benchmark | 51.3%† | 67.6% |
| GPQA Diamond (HELM) | HELM Capabilities | – | 75.3% |
| ARC-AGI-1 | ARC Prize | 96.5% | 60.8%† |
| ARC-AGI-2 | ARC Prize | 83.9% | 6.5%† |
| ARC-AGI-3 | ARC Prize | 0.8% | – |
| CritPt | Artificial Analysis | 30.0% | 1.1% |
| SciPredict | Scale AI SEAL | – | 17.9% |
| Chess Puzzles | Epoch AI Benchmarking Hub | 47.0% | 36.0%† |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 35.0% | 29.0%† |
| CL-bench | Epoch AI Benchmarking Hub | – | 17.8%† |
| LMCA | Epoch AI Benchmarking Hub | 55.0% | 39.7%† |
| DTBench | Epoch AI Benchmarking Hub | 93.3% | 84.8%† |
| ForecastBench | ForecastBench | – | 64.6% |
| Coding | |||
| SWE-bench Verified (Epoch) | Epoch AI Benchmarking Hub | – | 62.3%† |
| SciCode | SciCode | 55.0% | – |
Show all 162 benchmarks
| Metric | Publisher | GPT-5.6 Terra (max) | o3 |
|---|---|---|---|
| Coding | |||
| WeirdML | WeirdML | 78.3%† | 52.4%† |
| FrontierCode | Cognition | 41.3%† | – |
| GSO-Bench | GSO-Bench | – | 8.8%† |
| Arena Coding | Arena (LMArena) | 1517† | 1460 |
| Arena WebDev | Arena (LMArena) | 1520† | – |
| LiveBench 2.0 Coding | LiveBench | 86.2%† | – |
| LiveBench 1.0 Coding | LiveBench | 78.3% | – |
| LiveCodeBench | Vals AI | 85.9%† | 83.9%† |
| IOI | Vals AI | 87.6% | – |
| SWE-bench (Vals) | Vals AI | 95.4% | – |
| Aider Polyglot | Aider polyglot leaderboard | – | 81.3% |
| SWE-bench Verified (bash only) | SWE-bench | – | 58.4% |
| SWE-bench Verified (any scaffold) | SWE-bench | – | 58.4% |
| Code Migration | Vals AI | 47.8% | – |
| ProgramBench | Vals AI | 0.5% | – |
| Vibe Code Bench 1-100 | Vals AI | 14.8% | – |
| Vibe Code Bench v1.1 | Vals AI | 74.6% | – |
| CadEval | Epoch AI Benchmarking Hub | – | 74.0%† |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | 83.8%† | – |
| DeepSWE v1.1 | Epoch AI Benchmarking Hub | 69.6% | – |
| CursorBench | Cursor | 41.3% | – |
| ALE-Bench | Epoch AI Benchmarking Hub | 1951.4 | 933.5† |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 35.4% | – |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 88.0% | – |
| Startrise Frontend | Startrise | 66.9%† | – |
| GameBench 2 Overall | DuelLab GameBench 2 | 47.2† | 36.2 |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | 54.8† | 32.0† |
| LiveCodeBench (AA) | Artificial Analysis | – | 80.8% |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 21.5% | – |
| GDPval | OpenAI | – | 30.8%† |
| APEX-Agents | Mercor | 58.2% | – |
| Arena Agent | Arena (LMArena) | -1.4† | – |
| LiveBench Long Agentic | LiveBench | 33.1%† | – |
| LiveBench Browser | LiveBench | 70.3%† | – |
| LiveBench Agentic Coding | LiveBench | 55.0% | – |
| Terminal-Bench Hard | Artificial Analysis | 57.6% | 37.1% |
| τ²-Bench Telecom (AA) | Artificial Analysis | 86.3% | 80.7% |
| Terminal-Bench 2.1 (Vals) | Vals AI | 77.5% | – |
| BFCL Overall | Berkeley Function Calling Leaderboard | – | 63.0% |
| GDPval-AA v2.1, normalised score | Artificial Analysis | 47.6% | – |
| τ³-Banking (AA) | Artificial Analysis | 40.2% | – |
| ITBench SRE (AA) | Artificial Analysis | 51.0% | – |
| APEX-Agents (AA) | Artificial Analysis | 38.9% | – |
| CyberBench | Vals AI | 72.1% | – |
| Finance Agent v2 | Vals AI | 54.4% | – |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 0.8% | – |
| Legal Research Bench | Vals AI | 41.4% | – |
| SkillsBench | Vals AI | 58.9% | – |
| SREBench | Vals AI | 9.5% | – |
| Tax Agent Bench | Vals AI | 30.0% | – |
| Terminal-Bench 4.0 (Vals) | Vals AI | 22.7% | – |
| Terminal-Bench Science | Vals AI | 10.0% | – |
| BALROG | BALROG | 53.2% | – |
| DeepResearch Bench | DeepResearch Bench | – | 45.2%† |
| METR Time Horizons | METR | – | 63.6% |
| Vending-Bench 2 | Andon Labs | 7343.2† | – |
| AutomationBench | Artificial Analysis | 59.6% | – |
| GDP.pdf (AA) | Artificial Analysis | 24.0% | – |
| EnterpriseOps-Gym | Artificial Analysis | 38.5% | – |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 13.8% | – |
| BrowseComp-Plus | BrowseComp-Plus | – | 50.5% |
| FACTS Search | 81.7%† | – | |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 86.0% | 33.3%† |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 70.7% | – |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 99.7% | 83.9%† |
| MATH Level 5 | Epoch AI Benchmarking Hub | – | 97.8%† |
| ProofBench | Vals AI | 74.0%† | – |
| LiveBench Mathematics | LiveBench | 94.9% | – |
| AIME (Vals) | Vals AI | – | 85.3%† |
| MGSM | Vals AI | – | 91.8%† |
| IMO 2025 | MathArena | – | 16.7%† |
| Omni-MATH (HELM) | HELM Capabilities | – | 71.4% |
| MATH 500 | Vals AI | – | 94.6%† |
| Arena Maths | Arena (LMArena) | 1480† | 1448 |
| AIME 2025 (AA) | Artificial Analysis | – | 88.3% |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 43.2% | 49.4%† |
| LiveBench 2.0 Data Analysis | LiveBench | 79.3%† | – |
| LiveBench Research | LiveBench | 60.4%† | – |
| LiveBench 1.0 Data Analysis | LiveBench | 79.3% | – |
| AA-Omniscience v1.1 | Artificial Analysis | -9.3 | – |
| MMLU-Pro | Vals AI | 86.7%† | 85.6%† |
| LegalBench | Vals AI | 85.1% | 83.8%† |
| CorpFin | Vals AI | 65.3%† | 59.7%† |
| TaxEval | Vals AI | 76.2%† | 74.6%† |
| MedQA | Vals AI | – | 96.1%† |
| PRBench Finance | Scale AI SEAL | – | 47.7% |
| PRBench Legal | Scale AI SEAL | – | 48.6% |
| MultiNRC | Scale AI SEAL | – | 45.5%† |
| MMLU-Pro (HELM) | HELM Capabilities | – | 85.9% |
| Excel Modeling Benchmark | Vals AI | 66.2% | – |
| MedCode | Vals AI | 43.4%† | 47.3%† |
| MedScribe | Vals AI | 82.9%† | 76.7%† |
| Public Benefits Bench v1.1 | Vals AI | 62.4% | – |
| GDP.pdf | Epoch AI Benchmarking Hub | 24.7%† | – |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 42.3% | – |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 10.6% | – |
| Instruction following | |||
| LiveBench Language | LiveBench | 82.9% | – |
| LiveBench Instruction Following | LiveBench | 64.6% | – |
| IFBench | Artificial Analysis | 71.2% | 71.4% |
| MultiChallenge | Scale AI SEAL | – | 56.6%† |
| TutorBench | Scale AI SEAL | – | 52.1%† |
| IFEval (HELM) | HELM Capabilities | – | 86.9% |
| WildBench (HELM) | HELM Capabilities | – | 86.1% |
| Arena Instruction Following | Arena (LMArena) | 1463† | 1403 |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1467† | 1432 |
| EQ-Bench 4 | EQ-Bench | 1220† | – |
| Arena Creative Writing | Arena (LMArena) | 1429† | 1383 |
| Arena Multi-turn | Arena (LMArena) | 1474† | 1420 |
| Arena Longer Queries | Arena (LMArena) | 1468† | 1411 |
| Lech Mazur Creative Writing | Lech Mazur | – | 83.9%† |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1271† | 1214 |
| LiveBench Design | LiveBench | 76.0%† | – |
| MMMU-Pro | Artificial Analysis | 80.7% | 70.1% |
| VISTA | Scale AI SEAL | – | 50.1%† |
| MMMU (validation) | MMMU | – | 82.9% |
| MMMU-Pro (official) | MMMU | – | 76.4% |
| MMMU-Pro (Vals) | Vals AI | 86.5% | 80.4%† |
| MortgageTax | Vals AI | 67.3%† | 65.7%† |
| SAGE | Vals AI | 87.0% | – |
| VTB | Scale AI SEAL | – | 13.7% |
| GeoBench | GeoBench | – | 60.0%† |
| VPCT | Epoch AI Benchmarking Hub | – | 52.0%† |
| Blueprint-Bench 2 | Epoch AI Benchmarking Hub | 30.8%† | – |
| Long context | |||
| Fiction.LiveBench 120k | Fiction.live | – | 100.0%† |
| AA-LCR v1.1 | Artificial Analysis | 83.0% | 74.7% |
| Arena Document | Arena (LMArena) | 1477† | – |
| MLCR | Artificial Analysis | 27.2% | – |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 160.1† | 146.9 |
| LiveBench 2.0 | LiveBench | 67.8%† | – |
| LiveBench 1.0 | LiveBench | 77.9% | – |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 42.1 | 20.2 |
| AA Progress Index v1 | Artificial Analysis | 323.9 | 258.3 |
| HELM Capabilities mean | HELM Capabilities | – | 81.1% |
| Vals Multimodal Index | Vals AI | 65.1%† | – |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | – | 16.0% |
| MASK | Scale AI SEAL | – | 84.5%† |
| PropensityBench | Scale AI SEAL | – | 10.5% |
Data as of 2026-11-02. Best configuration of each model; every score links to its source on the model pages.
DuelLab GameBench 2, release gb2-20261026-cb69502a12d9 (https://benchmarks.duellab.org/releases/gb2-20261026-cb69502a12d9/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
What changed
- GPT-5.6 Terra (max): 1 new result (MirrorCode).
- GPT-5.6 Terra (max): BenchLeader Index 62.5 → 61.2.
- GPT-5.6 Terra (max): 2 new results (AA Progress Index v1, HLE-Diamond (AA)); 4 score revisions.
- o3: 1 new result (AA Progress Index v1).
- GPT-5.6 Terra (max): 1 new result (Harvey's Legal Agent Benchmark v1).
- GPT-5.6 Terra (max): 1 score revision.
- o3: 1 score revision.
- GPT-5.6 Terra (max): 5 new results (AutomationBench, GDP.pdf (AA), MLCR…).
- GPT-5.6 Terra (max): 1 new result (SREBench); 1 score revision.
- GPT-5.6 Terra (max): 1 score revision.
Related comparisons
- GPT-5.6 Terra vs Claude Opus 5.5
- GPT-5.6 Terra vs Gemini 4 Carbon
- GPT-5.6 Terra vs GPT-6 Astra
- o3 vs Claude Opus 5.5
- o3 vs Gemini 4 Carbon
- o3 vs GPT-6 Astra
- o3 vs GPT-6.1 Sol
- GPT-5.6 Terra vs Claude Fable 5.1
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
GPT-5.6 Terra vs o3: questions
- Is GPT-5.6 Terra better than o3?
- GPT-5.6 Terra (max) and o3 are neck and neck on quality (61.1 vs 61.0, within the ±6.0 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 2 Nov 2026, so check the category scores for your use.
- GPT-5.6 Terra vs o3: which is cheaper, and will that change?
- o3 is cheaper today: $3.50 against $4.50 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is GPT-5.6 Terra or o3 better for coding?
- GPT-5.6 Terra scores higher in coding (65 vs 62 on the category index, where 50 is average).
- Which is faster, GPT-5.6 Terra or o3?
- o3 streams faster (160 vs 130 output tokens per second, Artificial Analysis 7-day median).
- Is GPT-5.6 Terra or o3 better for agents and tool use?
- o3 scores higher in agentic tasks (70 vs 59 on the category index, where 50 is average). On Terminal-Bench Hard, the top shared agents & tools benchmark, it is 57.6% for GPT-5.6 Terra against 37.1% for o3.
More questions (4)
- Which is better for long documents?
- GPT-5.6 Terra scores higher on long-context benchmarks (66 vs 62). GPT-5.6 Terra accepts more context: 1,050,000 against 200,000 tokens for o3.
- Which has lower latency?
- o3 answers first: 6.8 s against 148.7 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-5.6 Terra vs o3?
- Across the 9 benchmarks both have, GPT-5.6 Terra leads 8, o3 leads 1. Every overall index that rates both puts GPT-5.6 Terra ahead.
- When was GPT-5.6 Terra released?
- GPT-5.6 Terra was released on 9 Jul 2026.