# Llama-3.3-70B-Instruct vs Step 3.7 Flash

> Step 3.7 Flash is the stronger model overall, scoring 36.5 to 30.3 on the Noometry Index. Llama-3.3-70B-Instruct costs 2.7× less per token, which makes it the better buy when Step 3.7 Flash's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/llama-3-3-70b-instruct-vs-step-3-7-flash
- Last updated: 2026-10-10
- Shared benchmarks: 2

## Summary

- They share 2 benchmarks with published results for both. Llama-3.3-70B-Instruct scores higher in 0 categories and Step 3.7 Flash in 3 categories; 3 gaps are clear of the uncertainty.
- The widest gap is in math, where Step 3.7 Flash leads 40.9 to 14.3.
- The biggest single-benchmark swing is SciCode: 26% for Llama-3.3-70B-Instruct and 40% for Step 3.7 Flash.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $0.18 / $1.11 for Step 3.7 Flash.
- Step 3.7 Flash accepts more context: 256K tokens versus 128K.

## Snapshot

| | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| Provider | Meta | StepFun |
| Noometry Index | 30.3 | 36.5 |
| Rank | 283 | 207 |
| Context | 128K | 256K |
| Input $/M | $0.10 | $0.18 |
| Output $/M | $0.32 | $1.11 |
| Weights | Open | Open |

## Coding

- Llama-3.3-70B-Instruct: 30.6 (#284)
- Step 3.7 Flash: 39.4 (#147)

| Benchmark | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| SciCode | 26% | 40% |
| WeirdML | 14.4% | — |
| BigCodeBench Instruct | 46.9% | — |
| LiveBench Coding | 36.6% | — |
| LMArena Coding | 1268 | — |
| BigCodeBench Complete | 57.5% | — |
| ALE-Bench | — | 694.12 |

## Agentic & Tool Use

- Llama-3.3-70B-Instruct: 25.6 (#100)
- Step 3.7 Flash: —

| Benchmark | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| Berkeley Function Calling Leaderboard | 31.9% | — |
| BALROG | 23% | — |

## Reasoning

- Llama-3.3-70B-Instruct: 13.7 (#325)
- Step 3.7 Flash: 20.9 (#211)

| Benchmark | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| CritPt | 0% | 2.3% |
| SimpleBench | 19.9% | — |
| NYT Connections (extended) | — | 39.7% |
| LiveBench Reasoning | 50.8% | — |
| LMArena Hard Prompts | 1257 | — |
| DTBench | 59.5% | — |
| LiveBench Data Analysis | 49.5% | — |
| LMCA | 17.5% | — |
| Epoch Capabilities Index | 127.32 | — |
| ForecastBench | 58.6 | — |
| LiveBench | 50.2% | — |

## Math

- Llama-3.3-70B-Instruct: 14.3 (#292)
- Step 3.7 Flash: 40.9 (#74)

| Benchmark | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| MathArena Final-Answer Competitions | — | 68.5% |
| OTIS Mock AIME 2024-2025 | 5.1% | — |
| LiveBench Math | 42.2% | — |
| LMArena Math | 1268 | — |
| MATH Level 5 | 41.6% | — |

## Knowledge

- Llama-3.3-70B-Instruct: 30.5 (#220)
- Step 3.7 Flash: —

| Benchmark | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| GPQA Diamond | 47.4% | — |
| Confabulations | 22.8% | — |
| Vectara Hallucination Rate | 4.1% | — |
| LMArena Expert | 1225 | — |
| MMLU | 86.3% | — |

## Multilingual

- Llama-3.3-70B-Instruct: 40.1 (#215)
- Step 3.7 Flash: —

| Benchmark | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| LMArena Non-English | 1236 | — |
| LMArena Chinese | 1217 | — |
| LMArena French | 1280 | — |
| LMArena German | 1251 | — |
| LMArena Japanese | 1149 | — |
| LMArena Korean | 1142 | — |
| LMArena Russian | 1252 | — |
| LMArena Spanish | 1269 | — |

## Instruction Following

- Llama-3.3-70B-Instruct: 71.3 (#152)
- Step 3.7 Flash: —

| Benchmark | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| LiveBench Instruction Following | 82.7% | — |
| LMArena Instruction Following | 1242 | — |

## Long Context

- Llama-3.3-70B-Instruct: 26.4 (#290)
- Step 3.7 Flash: —

| Benchmark | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| Fiction.LiveBench | 33.3% | — |
| LMArena Longer Query | 1256 | — |

## Writing & Preference

- Llama-3.3-70B-Instruct: 47.8 (#203)
- Step 3.7 Flash: —

| Benchmark | Llama-3.3-70B-Instruct | Step 3.7 Flash |
|---|---|---|
| LMArena Text | 1274 | — |
| LMArena Creative Writing | 1250 | — |
| LMArena Multi-Turn | 1279 | — |
| LiveBench Language | 39.2% | — |

## FAQ

### Is Llama-3.3-70B-Instruct better than Step 3.7 Flash?

Step 3.7 Flash is the stronger model overall, scoring 36.5 to 30.3 on the Noometry Index. Llama-3.3-70B-Instruct costs 2.7× less per token, which makes it the better buy when Step 3.7 Flash's lead doesn't matter for your workload.

### Which is cheaper, Llama-3.3-70B-Instruct or Step 3.7 Flash?

Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; Step 3.7 Flash lists at $0.18 and $1.11.

### Is Llama-3.3-70B-Instruct or Step 3.7 Flash better for coding?

Step 3.7 Flash scores higher on coding benchmarks: 39.4 versus 30.6 in the Noometry coding category.

### Which has the bigger context window?

Step 3.7 Flash does, with 256K tokens against 128K.

### How many benchmarks do Llama-3.3-70B-Instruct and Step 3.7 Flash share?

2 benchmarks have published results for both models. Llama-3.3-70B-Instruct has 43 scored results on Noometry and Step 3.7 Flash has 5.