Model comparison
GPT-4 vs Granite 3.1 2b Instruct
Granite 3.1 2b Instruct is the stronger model overall, scoring 33.2 to 29.1 on the Noometry Index.
Last verified . 12 shared benchmarks.
Summary
- They share 12 benchmarks with published results for both. GPT-4 scores higher in 4 categories and Granite 3.1 2b Instruct in 4 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Granite 3.1 2b Instruct leads 33.1 to 10.8.
- Granite 3.1 2b Instruct has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Granite 3.1 2b Instruct | |
|---|---|---|
| Provider | OpenAI | IBM |
| Noometry Index | 29.1 | 33.2 |
| Released | 2023-03-14 | — |
| Weights | Proprietary | Open |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 12 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Granite 3.1 2b Instruct leads
GPT-4: 31.6 (#283), Granite 3.1 2b Instruct: 33.4 (#257)
| Benchmark | GPT-4 | Granite 3.1 2b Instruct |
|---|---|---|
| LMArena Coding | 1254 | 1149 |
| WeirdML | 12.4% | — |
| BigCodeBench Instruct | 46% | — |
| BigCodeBench Complete | 57.2% | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Granite 3.1 2b Instruct: —
| Benchmark | GPT-4 | Granite 3.1 2b Instruct |
|---|---|---|
| METR Time Horizons | 36.1% | — |
Reasoning Granite 3.1 2b Instruct leads
GPT-4: 17.8 (#289), Granite 3.1 2b Instruct: 22.0 (#209)
| Benchmark | GPT-4 | Granite 3.1 2b Instruct |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1138 |
| Chess Puzzles | 4% | — |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| Epoch Capabilities Index | 125.89 | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Granite 3.1 2b Instruct leads
GPT-4: 10.8 (#309), Granite 3.1 2b Instruct: 33.1 (#206)
| Benchmark | GPT-4 | Granite 3.1 2b Instruct |
|---|---|---|
| LMArena Math | 1269 | 1159 |
| OTIS Mock AIME 2024-2025 | 1.1% | — |
| MATH Level 5 | 23% | — |
| GSM8K | 92% | — |
Knowledge Granite 3.1 2b Instruct leads
GPT-4: 18.4 (#282), Granite 3.1 2b Instruct: 30.8 (#224)
| Benchmark | GPT-4 | Granite 3.1 2b Instruct |
|---|---|---|
| LMArena Expert | 1211 | 1131 |
| GPQA Diamond | 35.7% | — |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multilingual GPT-4 leads
GPT-4: 40.6 (#215), Granite 3.1 2b Instruct: 29.1 (#269)
| Benchmark | GPT-4 | Granite 3.1 2b Instruct |
|---|---|---|
| LMArena Non-English | 1246 | 1068 |
| LMArena Chinese | 1242 | 1139 |
| LMArena Russian | 1251 | 1063 |
| LMArena French | 1283 | — |
| LMArena German | 1251 | — |
| LMArena Japanese | 1209 | — |
| LMArena Korean | 1184 | — |
| LMArena Spanish | 1261 | — |
Instruction Following GPT-4 leads
GPT-4: 65.3 (#222), Granite 3.1 2b Instruct: 57.7 (#264)
| Benchmark | GPT-4 | Granite 3.1 2b Instruct |
|---|---|---|
| LMArena Instruction Following | 1241 | 1116 |
Long Context GPT-4 leads
GPT-4: 37.7 (#212), Granite 3.1 2b Instruct: 35.0 (#244)
| Benchmark | GPT-4 | Granite 3.1 2b Instruct |
|---|---|---|
| LMArena Longer Query | 1244 | 1155 |
Writing & Preference Too close to call
GPT-4: 34.9 (#268), Granite 3.1 2b Instruct: 34.1 (#274)
| Benchmark | GPT-4 | Granite 3.1 2b Instruct |
|---|---|---|
| LMArena Text | 1263 | 1127 |
| LMArena Creative Writing | 1244 | 1116 |
| LMArena Multi-Turn | 1257 | 1099 |
| EQ-Bench Creative Writing | 752 | — |
Frequently asked questions
Is GPT-4 better than Granite 3.1 2b Instruct?
Granite 3.1 2b Instruct is the stronger model overall, scoring 33.2 to 29.1 on the Noometry Index.
Is GPT-4 or Granite 3.1 2b Instruct better for coding?
Granite 3.1 2b Instruct scores higher on coding benchmarks: 33.4 versus 31.6 in the Noometry coding category.
How many benchmarks do GPT-4 and Granite 3.1 2b Instruct share?
12 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Granite 3.1 2b Instruct has 12.