Model comparison
Claude Opus 4.8 vs GPT-5.6 Terra
Claude Opus 4.8 is the stronger model overall, scoring 60.7 to 59.2 on the Noometry Index. GPT-5.6 Terra costs 2.2× less per token, which makes it the better buy when Claude Opus 4.8's lead doesn't matter for your workload.
Last verified . 50 shared benchmarks.
Summary
- They share 50 benchmarks with published results for both. Claude Opus 4.8 scores higher in 8 categories and GPT-5.6 Terra in 2 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where Claude Opus 4.8 leads 47.6 to 40.1.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 88.8% for Claude Opus 4.8 and 51.3% for GPT-5.6 Terra.
- GPT-5.6 Terra is cheaper at $2 / $12 per million input/output tokens, against $5 / $25 for Claude Opus 4.8.
- GPT-5.6 Terra accepts more context: 1.05M tokens versus 1M.
Side by side
| Claude Opus 4.8 | GPT-5.6 Terra | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 60.7 | 59.2 |
| Released | 2026-05-28 | 2026-07-09 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $5 | $2 |
| Output $ / M tokens | $25 | $12 |
| Results tracked | 65 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 4.8 leads
Claude Opus 4.8: 59.9 (#12), GPT-5.6 Terra: 57.7 (#19)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| DeepSWE | 59% | 69.6% |
| FrontierCode | 46.5% | 41.3% |
| LMArena WebDev | 1556 | 1522 |
| SciCode | 53.5% | 55% |
| WeirdML | 82.9% | 78.3% |
| LMArena Coding | 1490 | 1484 |
| ALE-Bench | 1,564 | 1,951 |
| CursorBench | — | 41.3% |
| GSO | 47.1% | — |
Agentic & Tool Use Claude Opus 4.8 leads
Claude Opus 4.8: 47.6 (#11), GPT-5.6 Terra: 40.1 (#25)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| APEX-Agents | 48.9% | 58.2% |
| GDP.pdf | 24% | 24.7% |
| Vending-Bench 2 | 5,787 | 7,343 |
| OSWorld 2.0 | 20.6% | — |
| Remote Labor Index | 8.3% | — |
| τ²-bench Banking | 39.7% | — |
| DeepResearch Bench | 50.2% | — |
| PostTrainBench | 33.8% | — |
| BALROG | — | 53.2% |
| GBAEval | 70.9% | — |
| LMArena Search | 1204 | — |
Reasoning Claude Opus 4.8 leads
Claude Opus 4.8: 64.7 (#16), GPT-5.6 Terra: 60.7 (#21)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| ARC-AGI-2 | 72.1% | 83.9% |
| SimpleBench | 64.8% | 48.9% |
| Kagi LLM Benchmark | 88.8% | 51.3% |
| NYT Connections (extended) | 91.1% | 78.4% |
| ARC-AGI-1 | 92.5% | 96.5% |
| CritPt | 20.9% | 30% |
| Chess Puzzles | 34% | 54% |
| LMArena Hard Prompts | 1482 | 1468 |
| Mystery Game Puzzles | 36% | 35% |
| DTBench | 94.9% | 93.3% |
| LMCA | 57.5% | 55% |
| Surface Evolver Bench | 87.5% | 83.8% |
| Epoch Capabilities Index | 158.21 | 159.62 |
| EnigmaEval | 23.5% | — |
| EBR-Bench | 28.6% | — |
| Bench to the Future 3 | 0.14 | — |
| ForecastBench | 59.9 | — |
Math GPT-5.6 Terra leads
Claude Opus 4.8: 78.4 (#13), GPT-5.6 Terra: 81.6 (#12)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| FrontierMath (Tiers 1-3) | 80% | 86% |
| FrontierMath Tier 4 | 56.1% | 70.7% |
| OTIS Mock AIME 2024-2025 | 98.3% | 99.7% |
| ProofBench | 69% | 74% |
| LMArena Math | 1487 | 1466 |
| MathArena Final-Answer Competitions | 91.8% | — |
| FrontierMath (Feb 2025 set) | 47.2% | — |
| FrontierMath Tier 4 (v1) | 31.3% | — |
Knowledge Too close to call
Claude Opus 4.8: 61.3 (#29), GPT-5.6 Terra: 61.2 (#30)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| GPQA Diamond | 91% | 93.3% |
| SimpleQA Verified | 53% | 43.2% |
| LMArena Expert | 1502 | 1492 |
Multimodal GPT-5.6 Terra leads
Claude Opus 4.8: 42.9 (#26), GPT-5.6 Terra: 47.3 (#11)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| LMArena Vision | 1294 | 1271 |
| Blueprint-Bench 2 | 14.5% | 30.8% |
| Furniture Assembly | 42.5% | 54.2% |
| LMArena Document | 1475 | 1472 |
Multilingual Too close to call
Claude Opus 4.8: 55.2 (#33), GPT-5.6 Terra: 54.4 (#44)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| LMArena Non-English | 1450 | 1439 |
| LMArena Chinese | 1507 | 1513 |
| LMArena French | 1481 | 1471 |
| LMArena German | 1472 | 1460 |
| LMArena Japanese | 1440 | 1457 |
| LMArena Korean | 1432 | 1425 |
| LMArena Russian | 1474 | 1450 |
| LMArena Spanish | 1466 | 1448 |
Instruction Following Too close to call
Claude Opus 4.8: 77.4 (#24), GPT-5.6 Terra: 76.4 (#40)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| LMArena Instruction Following | 1476 | 1454 |
Long Context Claude Opus 4.8 leads
Claude Opus 4.8: 45.4 (#35), GPT-5.6 Terra: 44.4 (#68)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| LMArena Longer Query | 1483 | 1451 |
Writing & Preference Claude Opus 4.8 leads
Claude Opus 4.8: 72.0 (#16), GPT-5.6 Terra: 70.2 (#23)
| Benchmark | Claude Opus 4.8 | GPT-5.6 Terra |
|---|---|---|
| LMArena Text | 1461 | 1447 |
| LMArena Creative Writing | 1454 | 1410 |
| EQ-Bench Creative Writing | 1840 | 1855 |
| EQ-Bench 4 | 1281 | 1234 |
| LMArena Multi-Turn | 1476 | 1449 |
Frequently asked questions
Is Claude Opus 4.8 better than GPT-5.6 Terra?
Claude Opus 4.8 is the stronger model overall, scoring 60.7 to 59.2 on the Noometry Index. GPT-5.6 Terra costs 2.2× less per token, which makes it the better buy when Claude Opus 4.8's lead doesn't matter for your workload.
Which is cheaper, Claude Opus 4.8 or GPT-5.6 Terra?
GPT-5.6 Terra is cheaper. It lists at $2 per million input tokens and $12 per million output tokens; Claude Opus 4.8 lists at $5 and $25.
Is Claude Opus 4.8 or GPT-5.6 Terra better for coding?
Claude Opus 4.8 scores higher on coding benchmarks: 59.9 versus 57.7 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Terra does, with 1.05M tokens against 1M.
How many benchmarks do Claude Opus 4.8 and GPT-5.6 Terra share?
50 benchmarks have published results for both models. Claude Opus 4.8 has 65 scored results on Noometry and GPT-5.6 Terra has 52.