Model comparison
Llama 3.1-8B vs Llama-3.3-70B-Instruct
Llama-3.3-70B-Instruct is the stronger model overall, scoring 30.3 to 22.8 on the Noometry Index. Llama 3.1-8B costs 2.7× less per token, which makes it the better buy when Llama-3.3-70B-Instruct's lead doesn't matter for your workload.
Last verified . 31 shared benchmarks.
MetaSummary
- They share 31 benchmarks with published results for both. Llama 3.1-8B scores higher in 2 categories and Llama-3.3-70B-Instruct in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Llama-3.3-70B-Instruct leads 30.5 to 8.0.
- The biggest single-benchmark swing is GPQA Diamond: 27% for Llama 3.1-8B and 47.4% for Llama-3.3-70B-Instruct.
- Llama 3.1-8B is cheaper at $0.05 / $0.08 per million input/output tokens, against $0.10 / $0.32 for Llama-3.3-70B-Instruct.
Side by side
| Llama 3.1-8B | Llama-3.3-70B-Instruct | |
|---|---|---|
| Provider | Meta | Meta |
| Noometry Index | 22.8 | 30.3 |
| Released | 2024-07-23 | 2024-12-06 |
| Weights | Open | Open |
| Context window | 128K | 128K |
| Max output | 4K | 4K |
| Input $ / M tokens | $0.05 | $0.10 |
| Output $ / M tokens | $0.08 | $0.32 |
| Results tracked | 43 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama-3.3-70B-Instruct leads
Llama 3.1-8B: 19.9 (#335), Llama-3.3-70B-Instruct: 30.6 (#284)
| Benchmark | Llama 3.1-8B | Llama-3.3-70B-Instruct |
|---|---|---|
| SciCode | 13.2% | 26% |
| WeirdML | 1.7% | 14.4% |
| BigCodeBench Instruct | 32.8% | 46.9% |
| LMArena Coding | 1195 | 1268 |
| BigCodeBench Complete | 40.5% | 57.5% |
| LiveBench Coding | — | 36.6% |
| HumanEval+ | 62.8% | — |
| MBPP+ | 55.6% | — |
Agentic & Tool Use Llama-3.3-70B-Instruct leads
Llama 3.1-8B: 22.3 (#123), Llama-3.3-70B-Instruct: 25.6 (#100)
| Benchmark | Llama 3.1-8B | Llama-3.3-70B-Instruct |
|---|---|---|
| Berkeley Function Calling Leaderboard | 25.8% | 31.9% |
| BALROG | 15.1% | 23% |
Reasoning Too close to call
Llama 3.1-8B: 14.5 (#316), Llama-3.3-70B-Instruct: 13.7 (#325)
| Benchmark | Llama 3.1-8B | Llama-3.3-70B-Instruct |
|---|---|---|
| CritPt | 0% | 0% |
| LMArena Hard Prompts | 1175 | 1257 |
| DTBench | 50.9% | 59.5% |
| LMCA | 5.4% | 17.5% |
| Epoch Capabilities Index | 116.48 | 127.32 |
| SimpleBench | — | 19.9% |
| Chess Puzzles | 0% | — |
| LiveBench Reasoning | — | 50.8% |
| LiveBench Data Analysis | — | 49.5% |
| ForecastBench | — | 58.6 |
| LiveBench | — | 50.2% |
| PIQA | 81.2% | — |
Math Llama-3.3-70B-Instruct leads
Llama 3.1-8B: 9.5 (#311), Llama-3.3-70B-Instruct: 14.3 (#292)
| Benchmark | Llama 3.1-8B | Llama-3.3-70B-Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.7% | 5.1% |
| LMArena Math | 1179 | 1268 |
| MATH Level 5 | 22.9% | 41.6% |
| Omni-MATH | 13.7% | — |
| LiveBench Math | — | 42.2% |
| GSM8K | 82.4% | — |
Knowledge Llama-3.3-70B-Instruct leads
Llama 3.1-8B: 8.0 (#302), Llama-3.3-70B-Instruct: 30.5 (#220)
| Benchmark | Llama 3.1-8B | Llama-3.3-70B-Instruct |
|---|---|---|
| GPQA Diamond | 27% | 47.4% |
| LMArena Expert | 1144 | 1225 |
| MMLU | 56.1% | 86.3% |
| MMLU-Pro | 40.6% | — |
| Confabulations | — | 22.8% |
| Vectara Hallucination Rate | — | 4.1% |
| GPQA (HELM) | 24.7% | — |
| BoolQ | 82.8% | — |
Multilingual Llama-3.3-70B-Instruct leads
Llama 3.1-8B: 34.2 (#244), Llama-3.3-70B-Instruct: 40.1 (#215)
| Benchmark | Llama 3.1-8B | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Non-English | 1148 | 1236 |
| LMArena Chinese | 1151 | 1217 |
| LMArena French | 1176 | 1280 |
| LMArena German | 1144 | 1251 |
| LMArena Japanese | 1061 | 1149 |
| LMArena Korean | 1053 | 1142 |
| LMArena Russian | 1158 | 1252 |
| LMArena Spanish | 1169 | 1269 |
Instruction Following Llama-3.3-70B-Instruct leads
Llama 3.1-8B: 59.1 (#253), Llama-3.3-70B-Instruct: 71.3 (#152)
| Benchmark | Llama 3.1-8B | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Instruction Following | 1159 | 1242 |
| LiveBench Instruction Following | — | 82.7% |
| IFEval | 74.3% | — |
Long Context Llama 3.1-8B leads
Llama 3.1-8B: 35.8 (#233), Llama-3.3-70B-Instruct: 26.4 (#290)
| Benchmark | Llama 3.1-8B | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Longer Query | 1182 | 1256 |
| Fiction.LiveBench | — | 33.3% |
Writing & Preference Llama-3.3-70B-Instruct leads
Llama 3.1-8B: 29.9 (#285), Llama-3.3-70B-Instruct: 47.8 (#203)
| Benchmark | Llama 3.1-8B | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Text | 1187 | 1274 |
| LMArena Creative Writing | 1154 | 1250 |
| LMArena Multi-Turn | 1172 | 1279 |
| EQ-Bench Creative Writing | 713 | — |
| WildBench | 68.7% | — |
| LiveBench Language | — | 39.2% |
Frequently asked questions
Is Llama 3.1-8B better than Llama-3.3-70B-Instruct?
Llama-3.3-70B-Instruct is the stronger model overall, scoring 30.3 to 22.8 on the Noometry Index. Llama 3.1-8B costs 2.7× less per token, which makes it the better buy when Llama-3.3-70B-Instruct's lead doesn't matter for your workload.
Which is cheaper, Llama 3.1-8B or Llama-3.3-70B-Instruct?
Llama 3.1-8B is cheaper. It lists at $0.05 per million input tokens and $0.08 per million output tokens; Llama-3.3-70B-Instruct lists at $0.10 and $0.32.
Is Llama 3.1-8B or Llama-3.3-70B-Instruct better for coding?
Llama-3.3-70B-Instruct scores higher on coding benchmarks: 30.6 versus 19.9 in the Noometry coding category.
Which has the bigger context window?
Both accept 128K tokens.
How many benchmarks do Llama 3.1-8B and Llama-3.3-70B-Instruct share?
31 benchmarks have published results for both models. Llama 3.1-8B has 43 scored results on Noometry and Llama-3.3-70B-Instruct has 43.