Model comparison
Llama-3.3-70B-Instruct vs Qwen3.8 Max
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 19× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Last verified . 24 shared benchmarks.
Meta
Alibaba (Qwen)Summary
- They share 24 benchmarks with published results for both. Llama-3.3-70B-Instruct scores higher in 0 categories and Qwen3.8 Max in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3.8 Max leads 73.2 to 15.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 5.1% for Llama-3.3-70B-Instruct and 100% for Qwen3.8 Max.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $2 / $6 for Qwen3.8 Max.
- Qwen3.8 Max accepts more context: 1M tokens versus 128K.
- Llama-3.3-70B-Instruct has downloadable open weights; the other is API-only.
Side by side
| Llama-3.3-70B-Instruct | Qwen3.8 Max | |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 30.6 | 56.8 |
| Released | 2024-12-06 | 2026-08-02 |
| Weights | Open | Proprietary |
| Context window | 128K | 1M |
| Max output | 4K | 131K |
| Input $ / M tokens | $0.10 | $2 |
| Output $ / M tokens | $0.32 | $6 |
| Results tracked | 43 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 Max leads
Llama-3.3-70B-Instruct: 31.0 (#290), Qwen3.8 Max: 53.5 (#29)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| SciCode | 26% | 53.2% |
| LMArena Coding | 1268 | 1502 |
| DeepSWE | — | 57.5% |
| LMArena WebDev | — | 1674 |
| FrontierSWE | — | 17.8% |
| WeirdML | 14.4% | — |
| BigCodeBench Instruct | 46.9% | — |
| LiveBench Coding | 36.6% | — |
| BigCodeBench Complete | 57.5% | — |
Agentic & Tool Use Qwen3.8 Max leads
Llama-3.3-70B-Instruct: 25.8 (#105), Qwen3.8 Max: 45.4 (#14)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| APEX-Agents | — | 63.3% |
| Berkeley Function Calling Leaderboard | 31.9% | — |
| τ²-bench Banking | — | 55.1% |
| BALROG | 23% | — |
| GDP.pdf | — | 23.2% |
Reasoning Qwen3.8 Max leads
Llama-3.3-70B-Instruct: 14.1 (#327), Qwen3.8 Max: 54.4 (#26)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| CritPt | 0% | 20% |
| LMArena Hard Prompts | 1257 | 1496 |
| DTBench | 59.5% | 92% |
| LMCA | 17.5% | 46.2% |
| Epoch Capabilities Index | 127.33 | 156.41 |
| SimpleBench | 19.9% | — |
| NYT Connections (extended) | — | 88.3% |
| Chess Puzzles | — | 40% |
| LiveBench Reasoning | 50.8% | — |
| Mystery Game Puzzles | — | 38% |
| LiveBench Data Analysis | 49.5% | — |
| ForecastBench | 58.6 | — |
| LiveBench | 50.2% | — |
Math Qwen3.8 Max leads
Llama-3.3-70B-Instruct: 15.3 (#298), Qwen3.8 Max: 73.2 (#20)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 5.1% | 100% |
| LMArena Math | 1267 | 1499 |
| FrontierMath (Tiers 1-3) | — | 74.7% |
| FrontierMath Tier 4 | — | 46.3% |
| ProofBench | — | 58% |
| LiveBench Math | 42.2% | — |
| MATH Level 5 | 41.6% | — |
Knowledge Qwen3.8 Max leads
Llama-3.3-70B-Instruct: 30.6 (#226), Qwen3.8 Max: 61.7 (#27)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| GPQA Diamond | 47.4% | 92.7% |
| LMArena Expert | 1225 | 1507 |
| SimpleQA Verified | — | 47.3% |
| Confabulations | 22.8% | — |
| Vectara Hallucination Rate | 4.1% | — |
| MMLU | 86.3% | — |
Multimodal Not comparable
Llama-3.3-70B-Instruct: —, Qwen3.8 Max: 37.2 (#75)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| LMArena Vision | — | 1314 |
| Furniture Assembly | — | 20% |
Multilingual Qwen3.8 Max leads
Llama-3.3-70B-Instruct: 39.9 (#220), Qwen3.8 Max: 56.7 (#18)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| LMArena Non-English | 1236 | 1472 |
| LMArena Chinese | 1217 | 1538 |
| LMArena French | 1281 | 1503 |
| LMArena German | 1251 | 1483 |
| LMArena Japanese | 1150 | 1467 |
| LMArena Korean | 1143 | 1461 |
| LMArena Russian | 1252 | 1481 |
| LMArena Spanish | 1270 | 1492 |
Instruction Following Qwen3.8 Max leads
Llama-3.3-70B-Instruct: 71.1 (#157), Qwen3.8 Max: 77.6 (#17)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| LMArena Instruction Following | 1242 | 1479 |
| LiveBench Instruction Following | 82.7% | — |
Long Context Qwen3.8 Max leads
Llama-3.3-70B-Instruct: 26.4 (#295), Qwen3.8 Max: 45.6 (#31)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| LMArena Longer Query | 1256 | 1489 |
| Fiction.LiveBench | 33.3% | — |
Writing & Preference Qwen3.8 Max leads
Llama-3.3-70B-Instruct: 47.6 (#207), Qwen3.8 Max: 67.1 (#30)
| Benchmark | Llama-3.3-70B-Instruct | Qwen3.8 Max |
|---|---|---|
| LMArena Text | 1274 | 1483 |
| LMArena Creative Writing | 1250 | 1479 |
| LMArena Multi-Turn | 1280 | 1489 |
| LiveBench Language | 39.2% | — |
Frequently asked questions
Is Llama-3.3-70B-Instruct better than Qwen3.8 Max?
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 19× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Which is cheaper, Llama-3.3-70B-Instruct or Qwen3.8 Max?
Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; Qwen3.8 Max lists at $2 and $6.
Is Llama-3.3-70B-Instruct or Qwen3.8 Max better for coding?
Qwen3.8 Max scores higher on coding benchmarks: 53.5 versus 31.0 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 Max does, with 1M tokens against 128K.
How many benchmarks do Llama-3.3-70B-Instruct and Qwen3.8 Max share?
24 benchmarks have published results for both models. Llama-3.3-70B-Instruct has 43 scored results on Noometry and Qwen3.8 Max has 39.