Model comparison
Dolly 2.0-12b vs GPT-4 Turbo
GPT-4 Turbo is the stronger model overall, scoring 30.5 to 25.5 on the Noometry Index.
Last verified . 11 shared benchmarks.
DatabricksSummary
- They share 11 benchmarks with published results for both. Dolly 2.0-12b scores higher in 1 category and GPT-4 Turbo in 5 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where GPT-4 Turbo leads 47.7 to 15.2.
- Dolly 2.0-12b has downloadable open weights; the other is API-only.
Side by side
| Dolly 2.0-12b | GPT-4 Turbo | |
|---|---|---|
| Provider | Databricks | OpenAI |
| Noometry Index | 25.5 | 30.5 |
| Released | 2023-04-11 | 2023-11-06 |
| Weights | Open | Proprietary |
| Context window | — | 128K |
| Max output | — | 4K |
| Input $ / M tokens | — | $10 |
| Output $ / M tokens | — | $30 |
| Results tracked | 17 | 36 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-4 Turbo leads
Dolly 2.0-12b: 23.4 (#332), GPT-4 Turbo: 33.8 (#249)
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| LMArena Coding | 776 | 1268 |
| WeirdML | — | 18% |
| BigCodeBench Instruct | — | 48.2% |
| BigCodeBench Complete | — | 58.2% |
| HumanEval+ | — | 86.6% |
| MBPP+ | — | 73.3% |
Agentic & Tool Use Not comparable
Dolly 2.0-12b: —, GPT-4 Turbo: —
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| METR Time Horizons | — | 36.7% |
Reasoning Too close to call
Dolly 2.0-12b: 15.3 (#316), GPT-4 Turbo: 15.3 (#317)
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| LMArena Hard Prompts | 804 | 1251 |
| Epoch Capabilities Index | 89.67 | 127.25 |
| SimpleBench | — | 25.1% |
| Chess Puzzles | — | 6% |
| DTBench | — | 61.6% |
| LMCA | — | 9.8% |
| ForecastBench | — | 59.4 |
| HellaSwag | 70.8% | — |
| PIQA | 75.4% | — |
| WinoGrande | 61.8% | — |
Math Dolly 2.0-12b leads
Dolly 2.0-12b: 27.3 (#251), GPT-4 Turbo: 9.0 (#322)
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| LMArena Math | 871 | 1272 |
| FrontierMath (Tiers 1-3) | — | 0.7% |
| OTIS Mock AIME 2024-2025 | — | 6.7% |
| MATH Level 5 | — | 46.7% |
Knowledge Not comparable
Dolly 2.0-12b: —, GPT-4 Turbo: 24.3 (#268)
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| MMLU | 26.2% | 81.3% |
| GPQA Diamond | — | 46.6% |
| Confabulations | — | 28.4% |
| LMArena Expert | — | 1223 |
| ARC (AI2) Challenge | 39.6% | — |
| BoolQ | 56.3% | — |
| OpenBookQA | 39.2% | — |
Multimodal Not comparable
Dolly 2.0-12b: —, GPT-4 Turbo: 30.6 (#110)
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| LMArena Vision | — | 1090 |
Multilingual GPT-4 Turbo leads
Dolly 2.0-12b: 17.4 (#296), GPT-4 Turbo: 40.5 (#216)
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| LMArena Non-English | 836 | 1245 |
| LMArena Chinese | 836 | 1242 |
| LMArena French | — | 1276 |
| LMArena German | — | 1259 |
| LMArena Japanese | — | 1194 |
| LMArena Korean | — | 1187 |
| LMArena Russian | — | 1259 |
| LMArena Spanish | — | 1260 |
Instruction Following GPT-4 Turbo leads
Dolly 2.0-12b: 38.7 (#304), GPT-4 Turbo: 65.8 (#216)
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| LMArena Instruction Following | 814 | 1249 |
Long Context Not comparable
Dolly 2.0-12b: —, GPT-4 Turbo: 38.0 (#206)
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| LMArena Longer Query | — | 1254 |
Writing & Preference GPT-4 Turbo leads
Dolly 2.0-12b: 15.2 (#311), GPT-4 Turbo: 47.7 (#206)
| Benchmark | Dolly 2.0-12b | GPT-4 Turbo |
|---|---|---|
| LMArena Text | 851 | 1272 |
| LMArena Creative Writing | 864 | 1269 |
| LMArena Multi-Turn | 740 | 1267 |
Frequently asked questions
Is Dolly 2.0-12b better than GPT-4 Turbo?
GPT-4 Turbo is the stronger model overall, scoring 30.5 to 25.5 on the Noometry Index.
Is Dolly 2.0-12b or GPT-4 Turbo better for coding?
GPT-4 Turbo scores higher on coding benchmarks: 33.8 versus 23.4 in the Noometry coding category.
How many benchmarks do Dolly 2.0-12b and GPT-4 Turbo share?
11 benchmarks have published results for both models. Dolly 2.0-12b has 17 scored results on Noometry and GPT-4 Turbo has 36.