Model comparison
Llama 3.1-70B vs Qwen3 235B-A22B
Qwen3 235B-A22B is the stronger model overall, scoring 43.5 to 29.6 on the Noometry Index. Llama 3.1-70B costs 3.1× less per token, which makes it the better buy when Qwen3 235B-A22B'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 3.1-70B scores higher in 1 category and Qwen3 235B-A22B in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3 235B-A22B leads 50.4 to 13.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 3.6% for Llama 3.1-70B and 86.7% for Qwen3 235B-A22B.
- Llama 3.1-70B is cheaper at $0.40 / $0.40 per million input/output tokens, against $0.70 / $2.80 for Qwen3 235B-A22B.
- Qwen3 235B-A22B accepts more context: 131K tokens versus 128K.
Side by side
| Llama 3.1-70B | Qwen3 235B-A22B | |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 29.6 | 43.5 |
| Released | 2024-07-23 | 2025-04 |
| Weights | Open | Open |
| Context window | 128K | 131K |
| Max output | 4K | 16K |
| Input $ / M tokens | $0.40 | $0.70 |
| Output $ / M tokens | $0.40 | $2.80 |
| Results tracked | 35 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3 235B-A22B leads
Llama 3.1-70B: 30.3 (#296), Qwen3 235B-A22B: 44.3 (#75)
| Benchmark | Llama 3.1-70B | Qwen3 235B-A22B |
|---|---|---|
| WeirdML | 9% | 41% |
| LMArena Coding | 1260 | 1445 |
| Aider Polyglot | — | 59.6% |
| SciCode | — | 42.4% |
| BigCodeBench Instruct | 46.1% | — |
| BigCodeBench Complete | 54.8% | — |
Agentic & Tool Use Qwen3 235B-A22B leads
Llama 3.1-70B: 25.1 (#112), Qwen3 235B-A22B: 33.9 (#51)
| Benchmark | Llama 3.1-70B | Qwen3 235B-A22B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 52.1% |
| TheAgentCompany | 6.9% | — |
| BALROG | 27.9% | — |
| Vending-Bench 2 | — | -11.34 |
Reasoning Llama 3.1-70B leads
Llama 3.1-70B: 21.6 (#220), Qwen3 235B-A22B: 15.7 (#311)
| Benchmark | Llama 3.1-70B | Qwen3 235B-A22B |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1433 |
| DTBench | 60% | 80.3% |
| LMCA | 14.8% | 29.3% |
| Epoch Capabilities Index | 125.92 | 143.85 |
| ARC-AGI-2 | — | 1.3% |
| SimpleBench | — | 31% |
| Kagi LLM Benchmark | — | 69.4% |
| ARC-AGI-1 | — | 11% |
| CritPt | — | 0% |
| Chess Puzzles | — | 12% |
| Mystery Game Puzzles | — | 9% |
| ForecastBench | — | 59.7 |
Math Qwen3 235B-A22B leads
Llama 3.1-70B: 13.5 (#304), Qwen3 235B-A22B: 50.4 (#57)
| Benchmark | Llama 3.1-70B | Qwen3 235B-A22B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 3.6% | 86.7% |
| Omni-MATH | 21% | 71.8% |
| LMArena Math | 1252 | 1432 |
| MATH Level 5 | 36.7% | 68.9% |
| FrontierMath (Feb 2025 set) | — | 8.5% |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge Qwen3 235B-A22B leads
Llama 3.1-70B: 24.2 (#269), Qwen3 235B-A22B: 49.6 (#73)
| Benchmark | Llama 3.1-70B | Qwen3 235B-A22B |
|---|---|---|
| GPQA Diamond | 44.2% | 80.1% |
| MMLU-Pro | 65.3% | 84.4% |
| GPQA (HELM) | 42.6% | 72.7% |
| LMArena Expert | 1209 | 1463 |
| SimpleQA Verified | — | 40.4% |
| Confabulations | — | 15.6% |
| Vectara Hallucination Rate | — | 9.3% |
| MMLU | 80.1% | — |
Multilingual Qwen3 235B-A22B leads
Llama 3.1-70B: 38.8 (#225), Qwen3 235B-A22B: 52.3 (#89)
| Benchmark | Llama 3.1-70B | Qwen3 235B-A22B |
|---|---|---|
| LMArena Non-English | 1219 | 1409 |
| LMArena Chinese | 1215 | 1481 |
| LMArena French | 1261 | 1445 |
| LMArena German | 1222 | 1433 |
| LMArena Japanese | 1132 | 1399 |
| LMArena Korean | 1140 | 1391 |
| LMArena Russian | 1234 | 1411 |
| LMArena Spanish | 1253 | 1430 |
Instruction Following Qwen3 235B-A22B leads
Llama 3.1-70B: 65.3 (#223), Qwen3 235B-A22B: 72.6 (#136)
| Benchmark | Llama 3.1-70B | Qwen3 235B-A22B |
|---|---|---|
| IFEval | 82.1% | 83.5% |
| LMArena Instruction Following | 1231 | 1408 |
Long Context Qwen3 235B-A22B leads
Llama 3.1-70B: 37.6 (#214), Qwen3 235B-A22B: 46.1 (#26)
| Benchmark | Llama 3.1-70B | Qwen3 235B-A22B |
|---|---|---|
| LMArena Longer Query | 1241 | 1426 |
| Fiction.LiveBench | — | 75% |
Writing & Preference Qwen3 235B-A22B leads
Llama 3.1-70B: 35.4 (#267), Qwen3 235B-A22B: 59.6 (#108)
| Benchmark | Llama 3.1-70B | Qwen3 235B-A22B |
|---|---|---|
| LMArena Text | 1261 | 1419 |
| LMArena Creative Writing | 1232 | 1384 |
| EQ-Bench Creative Writing | 784 | 1366 |
| WildBench | 75.8% | 86.6% |
| LMArena Multi-Turn | 1256 | 1432 |
| Short-Story Creative Writing | — | 83% |
Frequently asked questions
Is Llama 3.1-70B better than Qwen3 235B-A22B?
Qwen3 235B-A22B is the stronger model overall, scoring 43.5 to 29.6 on the Noometry Index. Llama 3.1-70B costs 3.1× less per token, which makes it the better buy when Qwen3 235B-A22B's lead doesn't matter for your workload.
Which is cheaper, Llama 3.1-70B or Qwen3 235B-A22B?
Llama 3.1-70B is cheaper. It lists at $0.40 per million input tokens and $0.40 per million output tokens; Qwen3 235B-A22B lists at $0.70 and $2.80.
Is Llama 3.1-70B or Qwen3 235B-A22B better for coding?
Qwen3 235B-A22B scores higher on coding benchmarks: 44.3 versus 30.3 in the Noometry coding category.
Which has the bigger context window?
Qwen3 235B-A22B does, with 131K tokens against 128K.
How many benchmarks do Llama 3.1-70B and Qwen3 235B-A22B share?
30 benchmarks have published results for both models. Llama 3.1-70B has 35 scored results on Noometry and Qwen3 235B-A22B has 49.