Model comparison
Llama 4 Scout vs Qwen2.5 72B Instruct
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.5 to 27.4 on the Noometry Index. Llama 4 Scout costs 16× less per token, which makes it the better buy when Qwen2.5 72B Instruct's lead doesn't matter for your workload.
Last verified . 30 shared benchmarks.
Meta
Alibaba (Qwen)Summary
- They share 30 benchmarks with published results for both. Llama 4 Scout scores higher in 5 categories and Qwen2.5 72B Instruct in 4 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in coding, where Qwen2.5 72B Instruct leads 32.8 to 19.9.
- The biggest single-benchmark swing is BigCodeBench Complete: 43.1% for Llama 4 Scout and 55.9% for Qwen2.5 72B Instruct.
- Llama 4 Scout is cheaper at $0.10 / $0.30 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
- Qwen2.5 72B Instruct accepts more context: 131K tokens versus 128K.
Side by side
| Llama 4 Scout | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 27.4 | 31.5 |
| Released | 2025-04-05 | 2024-09 |
| Weights | Open | Open |
| Context window | 128K | 131K |
| Max output | 4K | 8K |
| Input $ / M tokens | $0.10 | $1.40 |
| Output $ / M tokens | $0.30 | $5.60 |
| Results tracked | 43 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen2.5 72B Instruct leads
Llama 4 Scout: 19.9 (#334), Qwen2.5 72B Instruct: 32.8 (#256)
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Coding | 1286 | 1293 |
| BigCodeBench Complete | 43.1% | 55.9% |
| SWE-bench Verified (bash only) | 9.1% | — |
| SciCode | 17% | — |
| WeirdML | — | 16% |
| BigCodeBench Instruct | — | 45.8% |
Agentic & Tool Use Llama 4 Scout leads
Llama 4 Scout: 24.4 (#109), Qwen2.5 72B Instruct: 22.0 (#124)
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| Berkeley Function Calling Leaderboard | 28.1% | — |
| TheAgentCompany | — | 5.7% |
| BALROG | — | 16.2% |
| METR Time Horizons | — | 35.8% |
Reasoning Qwen2.5 72B Instruct leads
Llama 4 Scout: 9.2 (#339), Qwen2.5 72B Instruct: 21.4 (#197)
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1266 | 1271 |
| DTBench | 57.9% | 62.9% |
| LMCA | 12% | 13.4% |
| Epoch Capabilities Index | 129.64 | 129.01 |
| ForecastBench | 57.5 | 57.5 |
| ARC-AGI-2 | 0% | — |
| Kagi LLM Benchmark | 36.9% | — |
| ARC-AGI-1 | 0.5% | — |
| CritPt | 0% | — |
| BIG-Bench Hard | — | 79.8% |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math Too close to call
Llama 4 Scout: 18.4 (#280), Qwen2.5 72B Instruct: 18.1 (#281)
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 7.8% | 8.1% |
| Omni-MATH | 37.3% | 33% |
| LMArena Math | 1286 | 1283 |
| MATH Level 5 | 62.3% | 63.2% |
| FrontierMath (Feb 2025 set) | 0% | — |
Knowledge Llama 4 Scout leads
Llama 4 Scout: 31.8 (#208), Qwen2.5 72B Instruct: 26.9 (#247)
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 51.8% | 49.1% |
| MMLU-Pro | 74.2% | 63.1% |
| GPQA (HELM) | 50.7% | 42.6% |
| LMArena Expert | 1236 | 1245 |
| Confabulations | — | 19.1% |
| Vectara Hallucination Rate | 7.7% | — |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multimodal Not comparable
Llama 4 Scout: 31.6 (#98), Qwen2.5 72B Instruct: —
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Vision | 1118 | — |
| SpatialViz-Bench | 34.2% | — |
Multilingual Too close to call
Llama 4 Scout: 41.3 (#207), Qwen2.5 72B Instruct: 41.2 (#208)
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1253 | 1252 |
| LMArena Chinese | 1256 | 1272 |
| LMArena French | 1283 | 1279 |
| LMArena German | 1274 | 1234 |
| LMArena Japanese | 1209 | 1180 |
| LMArena Korean | 1207 | 1188 |
| LMArena Russian | 1263 | 1264 |
| LMArena Spanish | 1277 | 1255 |
Instruction Following Too close to call
Llama 4 Scout: 66.0 (#212), Qwen2.5 72B Instruct: 65.7 (#216)
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| IFEval | 81.8% | 80.6% |
| LMArena Instruction Following | 1248 | 1254 |
Long Context Qwen2.5 72B Instruct leads
Llama 4 Scout: 27.5 (#289), Qwen2.5 72B Instruct: 38.9 (#183)
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1265 | 1282 |
| Fiction.LiveBench | 36% | — |
Writing & Preference Qwen2.5 72B Instruct leads
Llama 4 Scout: 37.1 (#256), Qwen2.5 72B Instruct: 46.9 (#210)
| Benchmark | Llama 4 Scout | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1279 | 1269 |
| LMArena Creative Writing | 1249 | 1221 |
| WildBench | 78% | 80.2% |
| LMArena Multi-Turn | 1281 | 1272 |
| EQ-Bench Creative Writing | 783 | — |
Frequently asked questions
Is Llama 4 Scout better than Qwen2.5 72B Instruct?
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.5 to 27.4 on the Noometry Index. Llama 4 Scout costs 16× less per token, which makes it the better buy when Qwen2.5 72B Instruct's lead doesn't matter for your workload.
Which is cheaper, Llama 4 Scout or Qwen2.5 72B Instruct?
Llama 4 Scout is cheaper. It lists at $0.10 per million input tokens and $0.30 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is Llama 4 Scout or Qwen2.5 72B Instruct better for coding?
Qwen2.5 72B Instruct scores higher on coding benchmarks: 32.8 versus 19.9 in the Noometry coding category.
Which has the bigger context window?
Qwen2.5 72B Instruct does, with 131K tokens against 128K.
How many benchmarks do Llama 4 Scout and Qwen2.5 72B Instruct share?
30 benchmarks have published results for both models. Llama 4 Scout has 43 scored results on Noometry and Qwen2.5 72B Instruct has 43.