# GPT-5.1 vs Llama 3-8B

> GPT-5.1 is the stronger model overall, scoring 48.8 to 25.3 on the Noometry Index. Llama 3-8B costs 25× less per token, which makes it the better buy when GPT-5.1's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gpt-5-1-vs-llama-3-8b
- Last updated: 2026-10-10
- Shared benchmarks: 23

## Summary

- They share 23 benchmarks with published results for both. GPT-5.1 scores higher in 8 categories and Llama 3-8B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-5.1 leads 50.5 to 7.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 88.6% for GPT-5.1 and 1.9% for Llama 3-8B.
- Llama 3-8B is cheaper at $0.14 / $0.14 per million input/output tokens, against $1.25 / $10 for GPT-5.1.
- GPT-5.1 accepts more context: 400K tokens versus 8K.
- Llama 3-8B has downloadable open weights; the other is API-only.

## Snapshot

| | GPT-5.1 | Llama 3-8B |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 48.8 | 25.3 |
| Rank | 47 | 338 |
| Context | 400K | 8K |
| Input $/M | $1.25 | $0.14 |
| Output $/M | $10 | $0.14 |
| Weights | Proprietary | Open |

## Coding

- GPT-5.1: 46.9 (#59)
- Llama 3-8B: 30.5 (#285)

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| LMArena Coding | 1454 | 1152 |
| SWE-bench Verified | 68% | — |
| SWE-bench Verified (bash only) | 66% | — |
| LMArena WebDev | 1395 | — |
| SciCode | 43.3% | — |
| GSO | 13.7% | — |
| WeirdML | 60.8% | — |
| BigCodeBench Instruct | — | 31.9% |
| LiveBench Coding | 72.5% | — |
| BigCodeBench Complete | — | 36.9% |
| ALE-Bench | 1,192 | — |
| HumanEval+ | — | 56.7% |
| MBPP+ | — | 54.8% |

## Agentic & Tool Use

- GPT-5.1: 32.5 (#55)
- Llama 3-8B: —

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| Terminal-Bench | 47.6% | — |
| DeepResearch Bench | 42.8% | — |
| LMArena Search | 1199 | — |
| Vending-Bench 2 | 1,473 | — |

## Reasoning

- GPT-5.1: 39.9 (#50)
- Llama 3-8B: 13.8 (#321)

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| Chess Puzzles | 32% | 0% |
| LMArena Hard Prompts | 1457 | 1133 |
| DTBench | 90.1% | 43.9% |
| Epoch Capabilities Index | 149.66 | 116.33 |
| ForecastBench | 58.1 | 58.6 |
| ARC-AGI-2 | 17.6% | — |
| SimpleBench | 53.2% | — |
| ARC-AGI-1 | 72.8% | — |
| CritPt | 4.9% | — |
| EnigmaEval | 11.2% | — |
| LiveBench Reasoning | 95.8% | — |
| Mystery Game Puzzles | 19% | — |
| LiveBench Data Analysis | 72.1% | — |
| LMCA | 43.9% | — |
| Adversarial NLI | — | 57.3% |
| LiveBench | 78.8% | — |
| WinoGrande | — | 75.7% |

## Math

- GPT-5.1: 50.3 (#45)
- Llama 3-8B: 8.3 (#317)

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.6% | 1.9% |
| LMArena Math | 1446 | 1151 |
| Omni-MATH | 46.4% | — |
| LiveBench Math | 94.5% | — |
| MATH Level 5 | — | 6.1% |
| FrontierMath (Feb 2025 set) | 31% | — |
| FrontierMath Tier 4 (v1) | 12.5% | — |

## Knowledge

- GPT-5.1: 50.5 (#65)
- Llama 3-8B: 7.8 (#303)

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| GPQA Diamond | 87.6% | 26.1% |
| LMArena Expert | 1470 | 1113 |
| Humanity's Last Exam | 23.7% | — |
| SimpleQA Verified | 48% | — |
| MMLU-Pro | 57.9% | — |
| Vectara Hallucination Rate | 10.9% | — |
| GPQA (HELM) | 44.2% | — |
| ARC (AI2) Challenge | — | 82.8% |
| MMLU | — | 68.8% |
| OpenBookQA | — | 82.6% |
| TriviaQA | — | 67.7% |

## Multimodal

- GPT-5.1: 44.0 (#17)
- Llama 3-8B: —

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| LMArena Vision | 1250 | — |
| VPCT | 58.7% | — |
| LMArena Document | 1403 | — |

## Multilingual

- GPT-5.1: 54.0 (#51)
- Llama 3-8B: 31.0 (#256)

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| LMArena Non-English | 1431 | 1098 |
| LMArena Chinese | 1495 | 1076 |
| LMArena French | 1451 | 1158 |
| LMArena German | 1439 | 1104 |
| LMArena Japanese | 1453 | 967 |
| LMArena Korean | 1401 | 1003 |
| LMArena Russian | 1435 | 1109 |
| LMArena Spanish | 1430 | 1172 |

## Instruction Following

- GPT-5.1: 84.0 (#1)
- Llama 3-8B: 58.7 (#255)

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| LMArena Instruction Following | 1442 | 1127 |
| LiveBench Instruction Following | 93.3% | — |
| IFEval | 93.5% | — |

## Long Context

- GPT-5.1: 47.6 (#14)
- Llama 3-8B: 34.2 (#246)

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| LMArena Longer Query | 1448 | 1128 |
| CL-bench | 23.7% | — |
| CL-bench Life | 17.3% | — |

## Writing & Preference

- GPT-5.1: 64.7 (#53)
- Llama 3-8B: 37.7 (#251)

| Benchmark | GPT-5.1 | Llama 3-8B |
|---|---|---|
| LMArena Text | 1443 | 1166 |
| LMArena Creative Writing | 1427 | 1150 |
| LMArena Multi-Turn | 1449 | 1152 |
| WildBench | 86.3% | — |
| LiveBench Language | 80.2% | — |

## FAQ

### Is GPT-5.1 better than Llama 3-8B?

GPT-5.1 is the stronger model overall, scoring 48.8 to 25.3 on the Noometry Index. Llama 3-8B costs 25× less per token, which makes it the better buy when GPT-5.1's lead doesn't matter for your workload.

### Which is cheaper, GPT-5.1 or Llama 3-8B?

Llama 3-8B is cheaper. It lists at $0.14 per million input tokens and $0.14 per million output tokens; GPT-5.1 lists at $1.25 and $10.

### Is GPT-5.1 or Llama 3-8B better for coding?

GPT-5.1 scores higher on coding benchmarks: 46.9 versus 30.5 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.1 does, with 400K tokens against 8K.

### How many benchmarks do GPT-5.1 and Llama 3-8B share?

23 benchmarks have published results for both models. GPT-5.1 has 63 scored results on Noometry and Llama 3-8B has 34.