Model comparison
Llama 3.1-8B vs MiMo-V2.5
MiMo-V2.5 is the stronger model overall, scoring 43.4 to 23.0 on the Noometry Index. Llama 3.1-8B costs 3.0× less per token, which makes it the better buy when MiMo-V2.5's lead doesn't matter for your workload.
Last verified . 19 shared benchmarks.
MetaSummary
- They share 19 benchmarks with published results for both. Llama 3.1-8B scores higher in 0 categories and MiMo-V2.5 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where MiMo-V2.5 leads 40.8 to 8.0.
- The biggest single-benchmark swing is SciCode: 13.2% for Llama 3.1-8B and 43.1% for MiMo-V2.5.
- Llama 3.1-8B is cheaper at $0.05 / $0.08 per million input/output tokens, against $0.14 / $0.28 for MiMo-V2.5.
- MiMo-V2.5 accepts more context: 1.05M tokens versus 128K.
Side by side
| Llama 3.1-8B | MiMo-V2.5 | |
|---|---|---|
| Provider | Meta | Xiaomi |
| Noometry Index | 23.0 | 43.4 |
| Released | 2024-07-23 | 2026-04-22 |
| Weights | Open | Open |
| Context window | 128K | 1.05M |
| Max output | 4K | 131K |
| Input $ / M tokens | $0.05 | $0.14 |
| Output $ / M tokens | $0.08 | $0.28 |
| Results tracked | 43 | 23 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding MiMo-V2.5 leads
Llama 3.1-8B: 20.2 (#340), MiMo-V2.5: 43.9 (#81)
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| SciCode | 13.2% | 43.1% |
| LMArena Coding | 1195 | 1469 |
| LMArena WebDev | — | 1438 |
| WeirdML | 1.7% | — |
| BigCodeBench Instruct | 32.8% | — |
| BigCodeBench Complete | 40.5% | — |
| ALE-Bench | — | 513.95 |
| HumanEval+ | 62.8% | — |
| MBPP+ | 55.6% | — |
Agentic & Tool Use Not comparable
Llama 3.1-8B: 22.5 (#131), MiMo-V2.5: —
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 25.8% | — |
| BALROG | 15.1% | — |
Reasoning MiMo-V2.5 leads
Llama 3.1-8B: 14.9 (#321), MiMo-V2.5: 28.6 (#101)
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| CritPt | 0% | 3.7% |
| LMArena Hard Prompts | 1175 | 1450 |
| Chess Puzzles | 0% | — |
| DTBench | 50.9% | — |
| LMCA | 5.4% | — |
| Epoch Capabilities Index | 116.57 | — |
| PIQA | 81.2% | — |
Math MiMo-V2.5 leads
Llama 3.1-8B: 10.2 (#317), MiMo-V2.5: 36.8 (#163)
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| LMArena Math | 1179 | 1436 |
| OTIS Mock AIME 2024-2025 | 1.7% | — |
| ProofBench | — | 16% |
| Omni-MATH | 13.7% | — |
| MATH Level 5 | 22.9% | — |
| GSM8K | 82.4% | — |
Knowledge MiMo-V2.5 leads
Llama 3.1-8B: 8.0 (#307), MiMo-V2.5: 40.8 (#115)
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| LMArena Expert | 1144 | 1460 |
| GPQA Diamond | 27% | — |
| MMLU-Pro | 40.6% | — |
| GPQA (HELM) | 24.7% | — |
| BoolQ | 82.8% | — |
| MMLU | 56.1% | — |
Multimodal Not comparable
Llama 3.1-8B: —, MiMo-V2.5: 39.8 (#54)
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| LMArena Vision | — | 1247 |
Multilingual MiMo-V2.5 leads
Llama 3.1-8B: 34.0 (#249), MiMo-V2.5: 51.9 (#99)
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| LMArena Non-English | 1148 | 1404 |
| LMArena Chinese | 1151 | 1468 |
| LMArena French | 1177 | 1447 |
| LMArena German | 1144 | 1421 |
| LMArena Japanese | 1061 | 1306 |
| LMArena Korean | 1053 | 1363 |
| LMArena Russian | 1158 | 1395 |
| LMArena Spanish | 1169 | 1416 |
Instruction Following MiMo-V2.5 leads
Llama 3.1-8B: 58.9 (#258), MiMo-V2.5: 75.5 (#60)
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| LMArena Instruction Following | 1159 | 1434 |
| IFEval | 74.3% | — |
Long Context MiMo-V2.5 leads
Llama 3.1-8B: 35.8 (#238), MiMo-V2.5: 44.2 (#73)
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| LMArena Longer Query | 1182 | 1445 |
Writing & Preference MiMo-V2.5 leads
Llama 3.1-8B: 29.7 (#290), MiMo-V2.5: 61.6 (#86)
| Benchmark | Llama 3.1-8B | MiMo-V2.5 |
|---|---|---|
| LMArena Text | 1187 | 1428 |
| LMArena Creative Writing | 1154 | 1393 |
| LMArena Multi-Turn | 1172 | 1445 |
| EQ-Bench Creative Writing | 713 | — |
| WildBench | 68.7% | — |
Frequently asked questions
Is Llama 3.1-8B better than MiMo-V2.5?
MiMo-V2.5 is the stronger model overall, scoring 43.4 to 23.0 on the Noometry Index. Llama 3.1-8B costs 3.0× less per token, which makes it the better buy when MiMo-V2.5's lead doesn't matter for your workload.
Which is cheaper, Llama 3.1-8B or MiMo-V2.5?
Llama 3.1-8B is cheaper. It lists at $0.05 per million input tokens and $0.08 per million output tokens; MiMo-V2.5 lists at $0.14 and $0.28.
Is Llama 3.1-8B or MiMo-V2.5 better for coding?
MiMo-V2.5 scores higher on coding benchmarks: 43.9 versus 20.2 in the Noometry coding category.
Which has the bigger context window?
MiMo-V2.5 does, with 1.05M tokens against 128K.
How many benchmarks do Llama 3.1-8B and MiMo-V2.5 share?
19 benchmarks have published results for both models. Llama 3.1-8B has 43 scored results on Noometry and MiMo-V2.5 has 23.