Model comparison
Llama 3.1-70B vs Ministral 8B
Llama 3.1-70B is the stronger model overall, scoring 29.6 to 28.2 on the Noometry Index. Ministral 8B costs 2.7× less per token, which makes it the better buy when Llama 3.1-70B's lead doesn't matter for your workload.
Last verified . 15 shared benchmarks.
Meta
Mistral AISummary
- They share 15 benchmarks with published results for both. Llama 3.1-70B scores higher in 6 categories and Ministral 8B in 3 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Ministral 8B leads 25.7 to 13.5.
- The biggest single-benchmark swing is MATH Level 5: 36.7% for Llama 3.1-70B and 14.9% for Ministral 8B.
- Ministral 8B is cheaper at $0.15 / $0.15 per million input/output tokens, against $0.40 / $0.40 for Llama 3.1-70B.
- Ministral 8B accepts more context: 262K tokens versus 128K.
Side by side
| Llama 3.1-70B | Ministral 8B | |
|---|---|---|
| Provider | Meta | Mistral AI |
| Noometry Index | 29.6 | 28.2 |
| Released | 2024-07-23 | 2024-10-01 |
| Weights | Open | Open |
| Context window | 128K | 262K |
| Max output | 4K | 262K |
| Input $ / M tokens | $0.40 | $0.15 |
| Output $ / M tokens | $0.40 | $0.15 |
| Results tracked | 35 | 17 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Ministral 8B leads
Llama 3.1-70B: 30.3 (#296), Ministral 8B: 35.0 (#230)
| Benchmark | Llama 3.1-70B | Ministral 8B |
|---|---|---|
| LMArena Coding | 1260 | 1202 |
| WeirdML | 9% | — |
| BigCodeBench Instruct | 46.1% | — |
| BigCodeBench Complete | 54.8% | — |
Agentic & Tool Use Llama 3.1-70B leads
Llama 3.1-70B: 25.1 (#112), Ministral 8B: 16.4 (#148)
| Benchmark | Llama 3.1-70B | Ministral 8B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 11.1% |
| TheAgentCompany | 6.9% | — |
| BALROG | 27.9% | — |
Reasoning Llama 3.1-70B leads
Llama 3.1-70B: 21.6 (#220), Ministral 8B: 18.4 (#281)
| Benchmark | Llama 3.1-70B | Ministral 8B |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1191 |
| DTBench | 60% | 45.7% |
| LMCA | 14.8% | — |
| Epoch Capabilities Index | 125.92 | — |
Math Ministral 8B leads
Llama 3.1-70B: 13.5 (#304), Ministral 8B: 25.7 (#267)
| Benchmark | Llama 3.1-70B | Ministral 8B |
|---|---|---|
| LMArena Math | 1252 | 1188 |
| MATH Level 5 | 36.7% | 14.9% |
| OTIS Mock AIME 2024-2025 | 3.6% | — |
| Omni-MATH | 21% | — |
Knowledge Llama 3.1-70B leads
Llama 3.1-70B: 24.2 (#269), Ministral 8B: 12.6 (#297)
| Benchmark | Llama 3.1-70B | Ministral 8B |
|---|---|---|
| GPQA Diamond | 44.2% | 27.1% |
| LMArena Expert | 1209 | 1170 |
| MMLU-Pro | 65.3% | — |
| Vectara Hallucination Rate | — | 7.4% |
| GPQA (HELM) | 42.6% | — |
| MMLU | 80.1% | — |
Multilingual Llama 3.1-70B leads
Llama 3.1-70B: 38.8 (#225), Ministral 8B: 35.1 (#247)
| Benchmark | Llama 3.1-70B | Ministral 8B |
|---|---|---|
| LMArena Non-English | 1219 | 1165 |
| LMArena Chinese | 1215 | 1193 |
| LMArena Russian | 1234 | 1195 |
| LMArena French | 1261 | — |
| LMArena German | 1222 | — |
| LMArena Japanese | 1132 | — |
| LMArena Korean | 1140 | — |
| LMArena Spanish | 1253 | — |
Instruction Following Llama 3.1-70B leads
Llama 3.1-70B: 65.3 (#223), Ministral 8B: 60.5 (#250)
| Benchmark | Llama 3.1-70B | Ministral 8B |
|---|---|---|
| LMArena Instruction Following | 1231 | 1161 |
| IFEval | 82.1% | — |
Long Context Too close to call
Llama 3.1-70B: 37.6 (#214), Ministral 8B: 36.7 (#227)
| Benchmark | Llama 3.1-70B | Ministral 8B |
|---|---|---|
| LMArena Longer Query | 1241 | 1212 |
Writing & Preference Ministral 8B leads
Llama 3.1-70B: 35.4 (#267), Ministral 8B: 39.6 (#246)
| Benchmark | Llama 3.1-70B | Ministral 8B |
|---|---|---|
| LMArena Text | 1261 | 1191 |
| LMArena Creative Writing | 1232 | 1175 |
| LMArena Multi-Turn | 1256 | 1166 |
| EQ-Bench Creative Writing | 784 | — |
| WildBench | 75.8% | — |
Frequently asked questions
Is Llama 3.1-70B better than Ministral 8B?
Llama 3.1-70B is the stronger model overall, scoring 29.6 to 28.2 on the Noometry Index. Ministral 8B costs 2.7× less per token, which makes it the better buy when Llama 3.1-70B's lead doesn't matter for your workload.
Which is cheaper, Llama 3.1-70B or Ministral 8B?
Ministral 8B is cheaper. It lists at $0.15 per million input tokens and $0.15 per million output tokens; Llama 3.1-70B lists at $0.40 and $0.40.
Is Llama 3.1-70B or Ministral 8B better for coding?
Ministral 8B scores higher on coding benchmarks: 35.0 versus 30.3 in the Noometry coding category.
Which has the bigger context window?
Ministral 8B does, with 262K tokens against 128K.
How many benchmarks do Llama 3.1-70B and Ministral 8B share?
15 benchmarks have published results for both models. Llama 3.1-70B has 35 scored results on Noometry and Ministral 8B has 17.