Model comparison
Llama-3.3-70B-Instruct vs Mistral Large
Mistral Large is the stronger model overall, scoring 31.9 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 19× less per token, which makes it the better buy when Mistral Large's lead doesn't matter for your workload.
Last verified . 40 shared benchmarks.
Meta
Mistral AISummary
- They share 40 benchmarks with published results for both. Llama-3.3-70B-Instruct scores higher in 3 categories and Mistral Large in 6 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in long context, where Mistral Large leads 38.3 to 26.4.
- The biggest single-benchmark swing is BigCodeBench Complete: 57.5% for Llama-3.3-70B-Instruct and 38.3% for Mistral Large.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $2 / $6 for Mistral Large.
- Mistral Large accepts more context: 131K tokens versus 128K.
Side by side
| Llama-3.3-70B-Instruct | Mistral Large | |
|---|---|---|
| Provider | Meta | Mistral AI |
| Noometry Index | 30.6 | 31.9 |
| Released | 2024-12-06 | 2024-02-26 |
| Weights | Open | Open |
| Context window | 128K | 131K |
| Max output | 4K | 16K |
| Input $ / M tokens | $0.10 | $2 |
| Output $ / M tokens | $0.32 | $6 |
| Results tracked | 43 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mistral Large leads
Llama-3.3-70B-Instruct: 31.0 (#290), Mistral Large: 34.3 (#240)
| Benchmark | Llama-3.3-70B-Instruct | Mistral Large |
|---|---|---|
| SciCode | 26% | 36.2% |
| BigCodeBench Instruct | 46.9% | 30% |
| LiveBench Coding | 36.6% | 47.1% |
| LMArena Coding | 1268 | 1277 |
| BigCodeBench Complete | 57.5% | 38.3% |
| WeirdML | 14.4% | — |
| ALE-Bench | — | 264.7 |
| HumanEval+ | — | 62.2% |
| MBPP+ | — | 59.5% |
Agentic & Tool Use Mistral Large leads
Llama-3.3-70B-Instruct: 25.8 (#105), Mistral Large: 28.6 (#89)
| Benchmark | Llama-3.3-70B-Instruct | Mistral Large |
|---|---|---|
| Berkeley Function Calling Leaderboard | 31.9% | 38.4% |
| BALROG | 23% | — |
Reasoning Mistral Large leads
Llama-3.3-70B-Instruct: 14.1 (#327), Mistral Large: 15.8 (#310)
| Benchmark | Llama-3.3-70B-Instruct | Mistral Large |
|---|---|---|
| SimpleBench | 19.9% | 22.5% |
| CritPt | 0% | 0% |
| LiveBench Reasoning | 50.8% | 43.5% |
| LMArena Hard Prompts | 1257 | 1257 |
| DTBench | 59.5% | 65.1% |
| LiveBench Data Analysis | 49.5% | 50.1% |
| LMCA | 17.5% | 16.7% |
| Epoch Capabilities Index | 127.33 | 128.52 |
| ForecastBench | 58.6 | 57.1 |
| LiveBench | 50.2% | 48.4% |
Math Mistral Large leads
Llama-3.3-70B-Instruct: 15.3 (#298), Mistral Large: 18.2 (#291)
| Benchmark | Llama-3.3-70B-Instruct | Mistral Large |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 5.1% | 8.5% |
| LiveBench Math | 42.2% | 42.5% |
| LMArena Math | 1267 | 1262 |
| MATH Level 5 | 41.6% | 50.3% |
| Omni-MATH | — | 28.1% |
| FrontierMath (Feb 2025 set) | — | 0.3% |
Knowledge Too close to call
Llama-3.3-70B-Instruct: 30.6 (#226), Mistral Large: 30.1 (#230)
| Benchmark | Llama-3.3-70B-Instruct | Mistral Large |
|---|---|---|
| GPQA Diamond | 47.4% | 51.3% |
| Confabulations | 22.8% | 21.4% |
| Vectara Hallucination Rate | 4.1% | 4.5% |
| LMArena Expert | 1225 | 1232 |
| MMLU | 86.3% | 80% |
| MMLU-Pro | — | 59.9% |
| GPQA (HELM) | — | 43.5% |
Multilingual Too close to call
Llama-3.3-70B-Instruct: 39.9 (#220), Mistral Large: 40.0 (#219)
| Benchmark | Llama-3.3-70B-Instruct | Mistral Large |
|---|---|---|
| LMArena Non-English | 1236 | 1237 |
| LMArena Chinese | 1217 | 1240 |
| LMArena French | 1281 | 1325 |
| LMArena German | 1251 | 1254 |
| LMArena Japanese | 1150 | 1188 |
| LMArena Korean | 1143 | 1202 |
| LMArena Russian | 1252 | 1257 |
| LMArena Spanish | 1270 | 1268 |
Instruction Following Llama-3.3-70B-Instruct leads
Llama-3.3-70B-Instruct: 71.1 (#157), Mistral Large: 67.9 (#191)
| Benchmark | Llama-3.3-70B-Instruct | Mistral Large |
|---|---|---|
| LiveBench Instruction Following | 82.7% | 67.9% |
| LMArena Instruction Following | 1242 | 1249 |
| IFEval | — | 87.7% |
Long Context Mistral Large leads
Llama-3.3-70B-Instruct: 26.4 (#295), Mistral Large: 38.3 (#199)
| Benchmark | Llama-3.3-70B-Instruct | Mistral Large |
|---|---|---|
| LMArena Longer Query | 1256 | 1261 |
| Fiction.LiveBench | 33.3% | — |
Writing & Preference Llama-3.3-70B-Instruct leads
Llama-3.3-70B-Instruct: 47.6 (#207), Mistral Large: 40.7 (#242)
| Benchmark | Llama-3.3-70B-Instruct | Mistral Large |
|---|---|---|
| LMArena Text | 1274 | 1266 |
| LMArena Creative Writing | 1250 | 1243 |
| LMArena Multi-Turn | 1280 | 1260 |
| LiveBench Language | 39.2% | 39.4% |
| Short-Story Creative Writing | — | 69% |
| EQ-Bench Creative Writing | — | 985 |
| WildBench | — | 80.1% |
Frequently asked questions
Is Llama-3.3-70B-Instruct better than Mistral Large?
Mistral Large is the stronger model overall, scoring 31.9 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 19× less per token, which makes it the better buy when Mistral Large's lead doesn't matter for your workload.
Which is cheaper, Llama-3.3-70B-Instruct or Mistral Large?
Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; Mistral Large lists at $2 and $6.
Is Llama-3.3-70B-Instruct or Mistral Large better for coding?
Mistral Large scores higher on coding benchmarks: 34.3 versus 31.0 in the Noometry coding category.
Which has the bigger context window?
Mistral Large does, with 131K tokens against 128K.
How many benchmarks do Llama-3.3-70B-Instruct and Mistral Large share?
40 benchmarks have published results for both models. Llama-3.3-70B-Instruct has 43 scored results on Noometry and Mistral Large has 51.