Model comparison
Llama 3.1-8B vs Qwen3.8 27B
Qwen3.8 27B is the stronger model overall, scoring 46.0 to 23.0 on the Noometry Index. Llama 3.1-8B costs 19× less per token, which makes it the better buy when Qwen3.8 27B'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 3.1-8B scores higher in 0 categories and Qwen3.8 27B in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Qwen3.8 27B leads 65.8 to 29.7.
- The biggest single-benchmark swing is DTBench: 50.9% for Llama 3.1-8B and 88% for Qwen3.8 27B.
- Llama 3.1-8B is cheaper at $0.05 / $0.08 per million input/output tokens, against $0.99 / $1.49 for Qwen3.8 27B.
- Qwen3.8 27B accepts more context: 262K tokens versus 128K.
Side by side
| Llama 3.1-8B | Qwen3.8 27B | |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 23.0 | 46.0 |
| Released | 2024-07-23 | 2026-08-14 |
| Weights | Open | Open |
| Context window | 128K | 262K |
| Max output | 4K | 33K |
| Input $ / M tokens | $0.05 | $0.99 |
| Output $ / M tokens | $0.08 | $1.49 |
| Results tracked | 43 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 27B leads
Llama 3.1-8B: 20.2 (#340), Qwen3.8 27B: 50.5 (#44)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| SciCode | 13.2% | 46.6% |
| LMArena Coding | 1195 | 1482 |
| LMArena WebDev | — | 1593 |
| WeirdML | 1.7% | — |
| BigCodeBench Instruct | 32.8% | — |
| BigCodeBench Complete | 40.5% | — |
| HumanEval+ | 62.8% | — |
| MBPP+ | 55.6% | — |
Agentic & Tool Use Qwen3.8 27B leads
Llama 3.1-8B: 22.5 (#131), Qwen3.8 27B: 32.9 (#57)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| APEX-Agents | — | 47.5% |
| Berkeley Function Calling Leaderboard | 25.8% | — |
| BALROG | 15.1% | — |
Reasoning Qwen3.8 27B leads
Llama 3.1-8B: 14.9 (#321), Qwen3.8 27B: 41.0 (#54)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| CritPt | 0% | 5.4% |
| LMArena Hard Prompts | 1175 | 1460 |
| DTBench | 50.9% | 88% |
| LMCA | 5.4% | 41.4% |
| Epoch Capabilities Index | 116.57 | 149.38 |
| ARC-AGI-2 | — | 42.4% |
| NYT Connections (extended) | — | 54.5% |
| ARC-AGI-1 | — | 87.5% |
| Chess Puzzles | 0% | — |
| Surface Evolver Bench | — | 45% |
| PIQA | 81.2% | — |
Math Qwen3.8 27B leads
Llama 3.1-8B: 10.2 (#317), Qwen3.8 27B: 37.1 (#161)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| LMArena Math | 1179 | 1456 |
| OTIS Mock AIME 2024-2025 | 1.7% | — |
| ProofBench | — | 16% |
| Omni-MATH | 13.7% | — |
| MATH Level 5 | 22.9% | — |
| GSM8K | 82.4% | — |
Knowledge Qwen3.8 27B leads
Llama 3.1-8B: 8.0 (#307), Qwen3.8 27B: 41.6 (#109)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| LMArena Expert | 1144 | 1482 |
| GPQA Diamond | 27% | — |
| MMLU-Pro | 40.6% | — |
| GPQA (HELM) | 24.7% | — |
| BoolQ | 82.8% | — |
| MMLU | 56.1% | — |
Multimodal Not comparable
Llama 3.1-8B: —, Qwen3.8 27B: 41.3 (#37)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| LMArena Vision | — | 1271 |
Multilingual Qwen3.8 27B leads
Llama 3.1-8B: 34.0 (#249), Qwen3.8 27B: 53.7 (#60)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| LMArena Non-English | 1148 | 1430 |
| LMArena Chinese | 1151 | 1504 |
| LMArena French | 1177 | 1465 |
| LMArena German | 1144 | 1438 |
| LMArena Japanese | 1061 | 1384 |
| LMArena Korean | 1053 | 1393 |
| LMArena Russian | 1158 | 1415 |
| LMArena Spanish | 1169 | 1448 |
Instruction Following Qwen3.8 27B leads
Llama 3.1-8B: 58.9 (#258), Qwen3.8 27B: 75.8 (#53)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| LMArena Instruction Following | 1159 | 1439 |
| IFEval | 74.3% | — |
Long Context Qwen3.8 27B leads
Llama 3.1-8B: 35.8 (#238), Qwen3.8 27B: 44.3 (#70)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| LMArena Longer Query | 1182 | 1450 |
Writing & Preference Qwen3.8 27B leads
Llama 3.1-8B: 29.7 (#290), Qwen3.8 27B: 65.8 (#43)
| Benchmark | Llama 3.1-8B | Qwen3.8 27B |
|---|---|---|
| LMArena Text | 1187 | 1441 |
| LMArena Creative Writing | 1154 | 1384 |
| EQ-Bench Creative Writing | 713 | 1671 |
| LMArena Multi-Turn | 1172 | 1441 |
| WildBench | 68.7% | — |
Frequently asked questions
Is Llama 3.1-8B better than Qwen3.8 27B?
Qwen3.8 27B is the stronger model overall, scoring 46.0 to 23.0 on the Noometry Index. Llama 3.1-8B costs 19× less per token, which makes it the better buy when Qwen3.8 27B's lead doesn't matter for your workload.
Which is cheaper, Llama 3.1-8B or Qwen3.8 27B?
Llama 3.1-8B is cheaper. It lists at $0.05 per million input tokens and $0.08 per million output tokens; Qwen3.8 27B lists at $0.99 and $1.49.
Is Llama 3.1-8B or Qwen3.8 27B better for coding?
Qwen3.8 27B scores higher on coding benchmarks: 50.5 versus 20.2 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 27B does, with 262K tokens against 128K.
How many benchmarks do Llama 3.1-8B and Qwen3.8 27B share?
23 benchmarks have published results for both models. Llama 3.1-8B has 43 scored results on Noometry and Qwen3.8 27B has 31.