Model comparison
Llama-3.3-70B-Instruct vs Qwen3 32B
Qwen3 32B is the stronger model overall, scoring 38.8 to 30.3 on the Noometry Index. Llama-3.3-70B-Instruct costs 7.9× less per token, which makes it the better buy when Qwen3 32B's lead doesn't matter for your workload.
Last verified . 23 shared benchmarks.
Meta
Alibaba (Qwen)Summary
- They share 23 benchmarks with published results for both. Llama-3.3-70B-Instruct scores higher in 1 category and Qwen3 32B in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3 32B leads 37.9 to 14.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 5.1% for Llama-3.3-70B-Instruct and 66.9% for Qwen3 32B.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $0.70 / $2.80 for Qwen3 32B.
- Qwen3 32B accepts more context: 131K tokens versus 128K.
Side by side
| Llama-3.3-70B-Instruct | Qwen3 32B | |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 30.3 | 38.8 |
| Released | 2024-12-06 | 2025-04 |
| Weights | Open | Open |
| Context window | 128K | 131K |
| Max output | 4K | 16K |
| Input $ / M tokens | $0.10 | $0.70 |
| Output $ / M tokens | $0.32 | $2.80 |
| Results tracked | 43 | 26 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3 32B leads
Llama-3.3-70B-Instruct: 30.6 (#284), Qwen3 32B: 37.3 (#184)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3 32B |
|---|---|---|
| SciCode | 26% | 35.4% |
| LMArena Coding | 1268 | 1358 |
| Aider Polyglot | — | 40% |
| WeirdML | 14.4% | — |
| BigCodeBench Instruct | 46.9% | — |
| LiveBench Coding | 36.6% | — |
| BigCodeBench Complete | 57.5% | — |
Agentic & Tool Use Qwen3 32B leads
Llama-3.3-70B-Instruct: 25.6 (#100), Qwen3 32B: 32.3 (#58)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3 32B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 31.9% | 48.7% |
| BALROG | 23% | — |
Reasoning Qwen3 32B leads
Llama-3.3-70B-Instruct: 13.7 (#325), Qwen3 32B: 19.6 (#233)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3 32B |
|---|---|---|
| CritPt | 0% | 0.3% |
| LMArena Hard Prompts | 1257 | 1335 |
| DTBench | 59.5% | 67.5% |
| LMCA | 17.5% | 17.3% |
| Epoch Capabilities Index | 127.32 | 138.51 |
| SimpleBench | 19.9% | — |
| Kagi LLM Benchmark | — | 54.9% |
| Chess Puzzles | — | 5% |
| LiveBench Reasoning | 50.8% | — |
| LiveBench Data Analysis | 49.5% | — |
| ForecastBench | 58.6 | — |
| LiveBench | 50.2% | — |
Math Qwen3 32B leads
Llama-3.3-70B-Instruct: 14.3 (#292), Qwen3 32B: 37.9 (#89)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3 32B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 5.1% | 66.9% |
| LMArena Math | 1268 | 1399 |
| LiveBench Math | 42.2% | — |
| MATH Level 5 | 41.6% | — |
Knowledge Qwen3 32B leads
Llama-3.3-70B-Instruct: 30.5 (#220), Qwen3 32B: 40.0 (#112)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3 32B |
|---|---|---|
| GPQA Diamond | 47.4% | 65.7% |
| Vectara Hallucination Rate | 4.1% | 5.9% |
| LMArena Expert | 1225 | 1362 |
| Confabulations | 22.8% | — |
| MMLU | 86.3% | — |
Multilingual Qwen3 32B leads
Llama-3.3-70B-Instruct: 40.1 (#215), Qwen3 32B: 45.8 (#162)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3 32B |
|---|---|---|
| LMArena Non-English | 1236 | 1317 |
| LMArena Chinese | 1217 | 1357 |
| LMArena German | 1251 | 1341 |
| LMArena Russian | 1252 | 1311 |
| LMArena French | 1280 | — |
| LMArena Japanese | 1149 | — |
| LMArena Korean | 1142 | — |
| LMArena Spanish | 1269 | — |
Instruction Following Llama-3.3-70B-Instruct leads
Llama-3.3-70B-Instruct: 71.3 (#152), Qwen3 32B: 69.1 (#174)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3 32B |
|---|---|---|
| LMArena Instruction Following | 1242 | 1305 |
| LiveBench Instruction Following | 82.7% | — |
Long Context Qwen3 32B leads
Llama-3.3-70B-Instruct: 26.4 (#290), Qwen3 32B: 43.7 (#83)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3 32B |
|---|---|---|
| Fiction.LiveBench | 33.3% | 74.2% |
| LMArena Longer Query | 1256 | 1327 |
Writing & Preference Qwen3 32B leads
Llama-3.3-70B-Instruct: 47.8 (#203), Qwen3 32B: 53.1 (#158)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3 32B |
|---|---|---|
| LMArena Text | 1274 | 1340 |
| LMArena Creative Writing | 1250 | 1297 |
| LMArena Multi-Turn | 1279 | 1331 |
| LiveBench Language | 39.2% | — |
Frequently asked questions
Is Llama-3.3-70B-Instruct better than Qwen3 32B?
Qwen3 32B is the stronger model overall, scoring 38.8 to 30.3 on the Noometry Index. Llama-3.3-70B-Instruct costs 7.9× less per token, which makes it the better buy when Qwen3 32B's lead doesn't matter for your workload.
Which is cheaper, Llama-3.3-70B-Instruct or Qwen3 32B?
Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; Qwen3 32B lists at $0.70 and $2.80.
Is Llama-3.3-70B-Instruct or Qwen3 32B better for coding?
Qwen3 32B scores higher on coding benchmarks: 37.3 versus 30.6 in the Noometry coding category.
Which has the bigger context window?
Qwen3 32B does, with 131K tokens against 128K.
How many benchmarks do Llama-3.3-70B-Instruct and Qwen3 32B share?
23 benchmarks have published results for both models. Llama-3.3-70B-Instruct has 43 scored results on Noometry and Qwen3 32B has 26.