Model comparison
Claude 3 Haiku vs Llama-3.3-70B-Instruct
Llama-3.3-70B-Instruct is the stronger model overall, scoring 30.6 to 25.9 on the Noometry Index.
Last verified . 29 shared benchmarks.
MetaSummary
- They share 29 benchmarks with published results for both. Claude 3 Haiku scores higher in 2 categories and Llama-3.3-70B-Instruct in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Llama-3.3-70B-Instruct leads 47.6 to 29.7.
- The biggest single-benchmark swing is MATH Level 5: 14.9% for Claude 3 Haiku and 41.6% for Llama-3.3-70B-Instruct.
- Llama-3.3-70B-Instruct has downloadable open weights; the other is API-only.
Side by side
| Claude 3 Haiku | Llama-3.3-70B-Instruct | |
|---|---|---|
| Provider | Anthropic | Meta |
| Noometry Index | 25.9 | 30.6 |
| Released | 2024-03-07 | 2024-12-06 |
| Weights | Proprietary | Open |
| Context window | — | 128K |
| Max output | — | 4K |
| Input $ / M tokens | — | $0.10 |
| Output $ / M tokens | — | $0.32 |
| Results tracked | 37 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama-3.3-70B-Instruct leads
Claude 3 Haiku: 26.4 (#325), Llama-3.3-70B-Instruct: 31.0 (#290)
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| WeirdML | 9.8% | 14.4% |
| BigCodeBench Instruct | 39.4% | 46.9% |
| LMArena Coding | 1199 | 1268 |
| BigCodeBench Complete | 50.1% | 57.5% |
| SciCode | — | 26% |
| LiveBench Coding | — | 36.6% |
| CadEval | 12% | — |
| HumanEval+ | 68.9% | — |
| MBPP+ | 68.8% | — |
Agentic & Tool Use Not comparable
Claude 3 Haiku: —, Llama-3.3-70B-Instruct: 25.8 (#105)
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 31.9% |
| BALROG | — | 23% |
Reasoning Claude 3 Haiku leads
Claude 3 Haiku: 16.3 (#307), Llama-3.3-70B-Instruct: 14.1 (#327)
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Hard Prompts | 1174 | 1257 |
| DTBench | 50.1% | 59.5% |
| LMCA | 8.8% | 17.5% |
| Epoch Capabilities Index | 118.35 | 127.33 |
| ForecastBench | 53.2 | 58.6 |
| SimpleBench | — | 19.9% |
| Kagi LLM Benchmark | 34.2% | — |
| CritPt | — | 0% |
| LiveBench Reasoning | — | 50.8% |
| LiveBench Data Analysis | — | 49.5% |
| LiveBench | — | 50.2% |
| WinoGrande | 74.2% | — |
Math Llama-3.3-70B-Instruct leads
Claude 3 Haiku: 9.8 (#319), Llama-3.3-70B-Instruct: 15.3 (#298)
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.8% | 5.1% |
| LMArena Math | 1188 | 1267 |
| MATH Level 5 | 14.9% | 41.6% |
| LiveBench Math | — | 42.2% |
Knowledge Llama-3.3-70B-Instruct leads
Claude 3 Haiku: 17.3 (#285), Llama-3.3-70B-Instruct: 30.6 (#226)
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| GPQA Diamond | 36.3% | 47.4% |
| Confabulations | 34.2% | 22.8% |
| LMArena Expert | 1148 | 1225 |
| MMLU | 73.8% | 86.3% |
| Vectara Hallucination Rate | — | 4.1% |
Multimodal Not comparable
Claude 3 Haiku: 23.6 (#128), Llama-3.3-70B-Instruct: —
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Vision | 950 | — |
| ScienceQA | 72% | — |
Multilingual Llama-3.3-70B-Instruct leads
Claude 3 Haiku: 36.0 (#243), Llama-3.3-70B-Instruct: 39.9 (#220)
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Non-English | 1178 | 1236 |
| LMArena Chinese | 1155 | 1217 |
| LMArena French | 1195 | 1281 |
| LMArena German | 1174 | 1251 |
| LMArena Japanese | 1102 | 1150 |
| LMArena Korean | 1109 | 1143 |
| LMArena Russian | 1204 | 1252 |
| LMArena Spanish | 1166 | 1270 |
Instruction Following Llama-3.3-70B-Instruct leads
Claude 3 Haiku: 61.3 (#247), Llama-3.3-70B-Instruct: 71.1 (#157)
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Instruction Following | 1173 | 1242 |
| LiveBench Instruction Following | — | 82.7% |
Long Context Claude 3 Haiku leads
Claude 3 Haiku: 36.1 (#237), Llama-3.3-70B-Instruct: 26.4 (#295)
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Longer Query | 1190 | 1256 |
| Fiction.LiveBench | — | 33.3% |
Writing & Preference Llama-3.3-70B-Instruct leads
Claude 3 Haiku: 29.7 (#291), Llama-3.3-70B-Instruct: 47.6 (#207)
| Benchmark | Claude 3 Haiku | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Text | 1195 | 1274 |
| LMArena Creative Writing | 1157 | 1250 |
| LMArena Multi-Turn | 1190 | 1280 |
| EQ-Bench Creative Writing | 717 | — |
| LiveBench Language | — | 39.2% |
Frequently asked questions
Is Claude 3 Haiku better than Llama-3.3-70B-Instruct?
Llama-3.3-70B-Instruct is the stronger model overall, scoring 30.6 to 25.9 on the Noometry Index.
Is Claude 3 Haiku or Llama-3.3-70B-Instruct better for coding?
Llama-3.3-70B-Instruct scores higher on coding benchmarks: 31.0 versus 26.4 in the Noometry coding category.
How many benchmarks do Claude 3 Haiku and Llama-3.3-70B-Instruct share?
29 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and Llama-3.3-70B-Instruct has 43.