Model comparison
DBRX vs Llama 3-8B
DBRX is the stronger model overall, scoring 29.4 to 25.5 on the Noometry Index.
Last verified . 21 shared benchmarks.
Databricks
MetaSummary
- They share 21 benchmarks with published results for both. DBRX scores higher in 4 categories and Llama 3-8B in 4 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where DBRX leads 24.3 to 8.8.
- The biggest single-benchmark swing is GPQA Diamond: 32.9% for DBRX and 26.1% for Llama 3-8B.
Side by side
| DBRX | Llama 3-8B | |
|---|---|---|
| Provider | Databricks | Meta |
| Noometry Index | 29.4 | 25.5 |
| Released | 2024-03-27 | 2024-04-18 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 21 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DBRX leads
DBRX: 32.9 (#266), Llama 3-8B: 31.0 (#289)
| Benchmark | DBRX | Llama 3-8B |
|---|---|---|
| LMArena Coding | 1132 | 1152 |
| HumanEval+ | 70.1% | 56.7% |
| MBPP+ | 55.8% | 54.8% |
| BigCodeBench Instruct | — | 31.9% |
| BigCodeBench Complete | — | 36.9% |
Reasoning DBRX leads
DBRX: 21.4 (#222), Llama 3-8B: 14.3 (#326)
| Benchmark | DBRX | Llama 3-8B |
|---|---|---|
| LMArena Hard Prompts | 1113 | 1133 |
| Chess Puzzles | — | 0% |
| DTBench | — | 43.9% |
| Adversarial NLI | — | 57.3% |
| Epoch Capabilities Index | — | 116.45 |
| ForecastBench | — | 58.6 |
| WinoGrande | — | 75.7% |
Math DBRX leads
DBRX: 24.3 (#269), Llama 3-8B: 8.8 (#323)
| Benchmark | DBRX | Llama 3-8B |
|---|---|---|
| LMArena Math | 1145 | 1151 |
| MATH Level 5 | 11.7% | 6.1% |
| OTIS Mock AIME 2024-2025 | — | 1.9% |
Knowledge DBRX leads
DBRX: 14.9 (#294), Llama 3-8B: 7.8 (#308)
| Benchmark | DBRX | Llama 3-8B |
|---|---|---|
| GPQA Diamond | 32.9% | 26.1% |
| LMArena Expert | 1076 | 1113 |
| ARC (AI2) Challenge | — | 82.8% |
| MMLU | — | 68.8% |
| OpenBookQA | — | 82.6% |
| TriviaQA | — | 67.7% |
Multilingual Llama 3-8B leads
DBRX: 29.3 (#268), Llama 3-8B: 30.8 (#261)
| Benchmark | DBRX | Llama 3-8B |
|---|---|---|
| LMArena Non-English | 1071 | 1098 |
| LMArena Chinese | 1068 | 1076 |
| LMArena French | 1096 | 1159 |
| LMArena German | 1057 | 1104 |
| LMArena Japanese | 990 | 967 |
| LMArena Korean | 993 | 1004 |
| LMArena Russian | 1078 | 1109 |
| LMArena Spanish | 1064 | 1173 |
Instruction Following Too close to call
DBRX: 57.5 (#270), Llama 3-8B: 58.4 (#260)
| Benchmark | DBRX | Llama 3-8B |
|---|---|---|
| LMArena Instruction Following | 1112 | 1127 |
Long Context Too close to call
DBRX: 33.7 (#258), Llama 3-8B: 34.2 (#251)
| Benchmark | DBRX | Llama 3-8B |
|---|---|---|
| LMArena Longer Query | 1112 | 1128 |
Writing & Preference Llama 3-8B leads
DBRX: 33.6 (#275), Llama 3-8B: 37.5 (#256)
| Benchmark | DBRX | Llama 3-8B |
|---|---|---|
| LMArena Text | 1119 | 1166 |
| LMArena Creative Writing | 1104 | 1150 |
| LMArena Multi-Turn | 1111 | 1152 |
Frequently asked questions
Is DBRX better than Llama 3-8B?
DBRX is the stronger model overall, scoring 29.4 to 25.5 on the Noometry Index.
Is DBRX or Llama 3-8B better for coding?
DBRX scores higher on coding benchmarks: 32.9 versus 31.0 in the Noometry coding category.
How many benchmarks do DBRX and Llama 3-8B share?
21 benchmarks have published results for both models. DBRX has 21 scored results on Noometry and Llama 3-8B has 34.