# GPT-4.1 vs Llama-3.3-70B-Instruct

> GPT-4.1 is the stronger model overall, scoring 35.7 to 30.3 on the Noometry Index. Llama-3.3-70B-Instruct costs 23× less per token, which makes it the better buy when GPT-4.1's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gpt-4-1-vs-llama-3-3-70b-instruct
- Last updated: 2026-10-10
- Shared benchmarks: 29

## Summary

- They share 29 benchmarks with published results for both. GPT-4.1 scores higher in 8 categories and Llama-3.3-70B-Instruct in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-4.1 leads 40.0 to 26.4.
- The biggest single-benchmark swing is MATH Level 5: 83% for GPT-4.1 and 41.6% for Llama-3.3-70B-Instruct.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $2 / $8 for GPT-4.1.
- GPT-4.1 accepts more context: 1.05M tokens versus 128K.
- Llama-3.3-70B-Instruct has downloadable open weights; the other is API-only.

## Snapshot

| | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 35.7 | 30.3 |
| Rank | 215 | 283 |
| Context | 1.05M | 128K |
| Input $/M | $2 | $0.10 |
| Output $/M | $8 | $0.32 |
| Weights | Proprietary | Open |

## Coding

- GPT-4.1: 34.2 (#228)
- Llama-3.3-70B-Instruct: 30.6 (#284)

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| WeirdML | 39% | 14.4% |
| LMArena Coding | 1391 | 1268 |
| SWE-bench Verified | 48.5% | — |
| SWE-bench Verified (bash only) | 39.6% | — |
| Aider Polyglot | 52.4% | — |
| SciCode | — | 26% |
| BigCodeBench Instruct | — | 46.9% |
| LiveBench Coding | — | 36.6% |
| BigCodeBench Complete | — | 57.5% |
| CadEval | 42% | — |
| ALE-Bench | 558.1 | — |

## Agentic & Tool Use

- GPT-4.1: 34.4 (#40)
- Llama-3.3-70B-Instruct: 25.6 (#100)

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| Berkeley Function Calling Leaderboard | 54% | 31.9% |
| BALROG | — | 23% |

## Reasoning

- GPT-4.1: 11.7 (#334)
- Llama-3.3-70B-Instruct: 13.7 (#325)

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| SimpleBench | 27% | 19.9% |
| LMArena Hard Prompts | 1384 | 1257 |
| DTBench | 68.3% | 59.5% |
| LMCA | 25.6% | 17.5% |
| Epoch Capabilities Index | 136.79 | 127.32 |
| ForecastBench | 61.5 | 58.6 |
| ARC-AGI-2 | 0.4% | — |
| Kagi LLM Benchmark | 52.3% | — |
| ARC-AGI-1 | 5.5% | — |
| CritPt | — | 0% |
| Chess Puzzles | 6% | — |
| EnigmaEval | 2.2% | — |
| LiveBench Reasoning | — | 50.8% |
| LiveBench Data Analysis | — | 49.5% |
| LiveBench | — | 50.2% |

## Math

- GPT-4.1: 21.4 (#269)
- Llama-3.3-70B-Instruct: 14.3 (#292)

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 38.3% | 5.1% |
| LMArena Math | 1369 | 1268 |
| MATH Level 5 | 83% | 41.6% |
| FrontierMath (Tiers 1-3) | 6% | — |
| Omni-MATH | 47.1% | — |
| LiveBench Math | — | 42.2% |
| FrontierMath (Feb 2025 set) | 5.5% | — |
| FrontierMath Tier 4 (v1) | 0% | — |

## Knowledge

- GPT-4.1: 37.0 (#152)
- Llama-3.3-70B-Instruct: 30.5 (#220)

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| GPQA Diamond | 66.9% | 47.4% |
| Vectara Hallucination Rate | 5.6% | 4.1% |
| LMArena Expert | 1364 | 1225 |
| Humanity's Last Exam | 5.4% | — |
| SimpleQA Verified | 31.1% | — |
| MMLU-Pro | 81.1% | — |
| Confabulations | — | 22.8% |
| GPQA (HELM) | 65.9% | — |
| MMLU | — | 86.3% |

## Multimodal

- GPT-4.1: 37.5 (#66)
- Llama-3.3-70B-Instruct: —

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Vision | 1210 | — |
| GeoBench | 72% | — |

## Multilingual

- GPT-4.1: 49.6 (#128)
- Llama-3.3-70B-Instruct: 40.1 (#215)

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Non-English | 1370 | 1236 |
| LMArena Chinese | 1382 | 1217 |
| LMArena French | 1383 | 1280 |
| LMArena German | 1380 | 1251 |
| LMArena Japanese | 1320 | 1149 |
| LMArena Korean | 1339 | 1142 |
| LMArena Russian | 1377 | 1252 |
| LMArena Spanish | 1375 | 1269 |

## Instruction Following

- GPT-4.1: 71.5 (#148)
- Llama-3.3-70B-Instruct: 71.3 (#152)

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Instruction Following | 1367 | 1242 |
| LiveBench Instruction Following | — | 82.7% |
| IFEval | 83.8% | — |

## Long Context

- GPT-4.1: 40.0 (#158)
- Llama-3.3-70B-Instruct: 26.4 (#290)

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| Fiction.LiveBench | 63.9% | 33.3% |
| LMArena Longer Query | 1385 | 1256 |

## Writing & Preference

- GPT-4.1: 57.8 (#120)
- Llama-3.3-70B-Instruct: 47.8 (#203)

| Benchmark | GPT-4.1 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Text | 1383 | 1274 |
| LMArena Creative Writing | 1363 | 1250 |
| LMArena Multi-Turn | 1398 | 1279 |
| EQ-Bench Creative Writing | 1420 | — |
| WildBench | 85.4% | — |
| LiveBench Language | — | 39.2% |

## FAQ

### Is GPT-4.1 better than Llama-3.3-70B-Instruct?

GPT-4.1 is the stronger model overall, scoring 35.7 to 30.3 on the Noometry Index. Llama-3.3-70B-Instruct costs 23× less per token, which makes it the better buy when GPT-4.1's lead doesn't matter for your workload.

### Which is cheaper, GPT-4.1 or Llama-3.3-70B-Instruct?

Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; GPT-4.1 lists at $2 and $8.

### Is GPT-4.1 or Llama-3.3-70B-Instruct better for coding?

GPT-4.1 scores higher on coding benchmarks: 34.2 versus 30.6 in the Noometry coding category.

### Which has the bigger context window?

GPT-4.1 does, with 1.05M tokens against 128K.

### How many benchmarks do GPT-4.1 and Llama-3.3-70B-Instruct share?

29 benchmarks have published results for both models. GPT-4.1 has 52 scored results on Noometry and Llama-3.3-70B-Instruct has 43.