# GPT-5.6 Terra vs Llama 3.2 90B

> GPT-5.6 Terra is the stronger model overall, scoring 59.2 to 27.5 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gpt-5-6-terra-vs-llama-3-2-90b
- Last updated: 2026-10-10
- Shared benchmarks: 5

## Summary

- They share 5 benchmarks with published results for both. GPT-5.6 Terra scores higher in 5 categories and Llama 3.2 90B in 0 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Terra leads 81.6 to 11.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 99.7% for GPT-5.6 Terra and 2.6% for Llama 3.2 90B.
- Llama 3.2 90B has downloadable open weights; the other is API-only.

## Snapshot

| | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 59.2 | 27.5 |
| Rank | 17 | 331 |
| Context | 1.05M | — |
| Input $/M | $2 | — |
| Output $/M | $12 | — |
| Weights | Proprietary | Open |

## Coding

- GPT-5.6 Terra: 57.7 (#19)
- Llama 3.2 90B: —

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| DeepSWE | 69.6% | — |
| FrontierCode | 41.3% | — |
| CursorBench | 41.3% | — |
| LMArena WebDev | 1522 | — |
| SciCode | 55% | — |
| WeirdML | 78.3% | — |
| LMArena Coding | 1484 | — |
| ALE-Bench | 1,951 | — |

## Agentic & Tool Use

- GPT-5.6 Terra: 40.1 (#25)
- Llama 3.2 90B: 30.0 (#80)

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| BALROG | 53.2% | 27.3% |
| APEX-Agents | 58.2% | — |
| GDP.pdf | 24.7% | — |
| Vending-Bench 2 | 7,343 | — |

## Reasoning

- GPT-5.6 Terra: 60.7 (#21)
- Llama 3.2 90B: 21.7 (#217)

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| Epoch Capabilities Index | 159.62 | 125.5 |
| ARC-AGI-2 | 83.9% | — |
| SimpleBench | 48.9% | — |
| Kagi LLM Benchmark | 51.3% | — |
| NYT Connections (extended) | 78.4% | — |
| ARC-AGI-1 | 96.5% | — |
| CritPt | 30% | — |
| Chess Puzzles | 54% | — |
| EnigmaEval | — | 0.4% |
| LMArena Hard Prompts | 1468 | — |
| Mystery Game Puzzles | 35% | — |
| DTBench | 93.3% | — |
| LMCA | 55% | — |
| Surface Evolver Bench | 83.8% | — |

## Math

- GPT-5.6 Terra: 81.6 (#12)
- Llama 3.2 90B: 11.1 (#308)

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 99.7% | 2.6% |
| FrontierMath (Tiers 1-3) | 86% | — |
| FrontierMath Tier 4 | 70.7% | — |
| ProofBench | 74% | — |
| LMArena Math | 1466 | — |
| MATH Level 5 | — | 39.4% |

## Knowledge

- GPT-5.6 Terra: 61.2 (#30)
- Llama 3.2 90B: 21.7 (#274)

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| GPQA Diamond | 93.3% | 41% |
| SimpleQA Verified | 43.2% | — |
| LMArena Expert | 1492 | — |
| MMLU | — | 80.3% |

## Multimodal

- GPT-5.6 Terra: 47.3 (#11)
- Llama 3.2 90B: 25.4 (#124)

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| LMArena Vision | 1271 | 1000 |
| GeoBench | — | 52% |
| Blueprint-Bench 2 | 30.8% | — |
| Furniture Assembly | 54.2% | — |
| LMArena Document | 1472 | — |

## Multilingual

- GPT-5.6 Terra: 54.4 (#44)
- Llama 3.2 90B: —

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| LMArena Non-English | 1439 | — |
| LMArena Chinese | 1513 | — |
| LMArena French | 1471 | — |
| LMArena German | 1460 | — |
| LMArena Japanese | 1457 | — |
| LMArena Korean | 1425 | — |
| LMArena Russian | 1450 | — |
| LMArena Spanish | 1448 | — |

## Instruction Following

- GPT-5.6 Terra: 76.4 (#40)
- Llama 3.2 90B: —

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| LMArena Instruction Following | 1454 | — |

## Long Context

- GPT-5.6 Terra: 44.4 (#68)
- Llama 3.2 90B: —

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| LMArena Longer Query | 1451 | — |

## Writing & Preference

- GPT-5.6 Terra: 70.2 (#23)
- Llama 3.2 90B: —

| Benchmark | GPT-5.6 Terra | Llama 3.2 90B |
|---|---|---|
| LMArena Text | 1447 | — |
| LMArena Creative Writing | 1410 | — |
| EQ-Bench Creative Writing | 1855 | — |
| EQ-Bench 4 | 1234 | — |
| LMArena Multi-Turn | 1449 | — |

## FAQ

### Is GPT-5.6 Terra better than Llama 3.2 90B?

GPT-5.6 Terra is the stronger model overall, scoring 59.2 to 27.5 on the Noometry Index.

### How many benchmarks do GPT-5.6 Terra and Llama 3.2 90B share?

5 benchmarks have published results for both models. GPT-5.6 Terra has 52 scored results on Noometry and Llama 3.2 90B has 9.