Model comparison
Codellama 70b Instruct vs GPT-4
Codellama 70b Instruct is the stronger model overall, scoring 33.7 to 29.1 on the Noometry Index.
Last verified . 7 shared benchmarks.
MetaSummary
- They share 7 benchmarks with published results for both. Codellama 70b Instruct scores higher in 2 categories and GPT-4 in 3 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in multilingual, where GPT-4 leads 40.6 to 24.8.
- The biggest single-benchmark swing is BigCodeBench Complete: 49.6% for Codellama 70b Instruct and 57.2% for GPT-4.
- Codellama 70b Instruct has downloadable open weights; the other is API-only.
Side by side
| Codellama 70b Instruct | GPT-4 | |
|---|---|---|
| Provider | Meta | OpenAI |
| Noometry Index | 33.7 | 29.1 |
| Released | — | 2023-03-14 |
| Weights | Open | Proprietary |
| Context window | — | 8K |
| Max output | — | 8K |
| Input $ / M tokens | — | $30 |
| Output $ / M tokens | — | $60 |
| Results tracked | 7 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Codellama 70b Instruct leads
Codellama 70b Instruct: 37.6 (#193), GPT-4: 31.6 (#283)
| Benchmark | Codellama 70b Instruct | GPT-4 |
|---|---|---|
| BigCodeBench Instruct | 40.7% | 46% |
| BigCodeBench Complete | 49.6% | 57.2% |
| HumanEval+ | 65.9% | 79.3% |
| WeirdML | — | 12.4% |
| LMArena Coding | — | 1254 |
Agentic & Tool Use Not comparable
Codellama 70b Instruct: —, GPT-4: —
| Benchmark | Codellama 70b Instruct | GPT-4 |
|---|---|---|
| METR Time Horizons | — | 36.1% |
Reasoning Codellama 70b Instruct leads
Codellama 70b Instruct: 20.1 (#242), GPT-4: 17.8 (#289)
| Benchmark | Codellama 70b Instruct | GPT-4 |
|---|---|---|
| LMArena Hard Prompts | 1052 | 1241 |
| 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 Not comparable
Codellama 70b Instruct: —, GPT-4: 10.8 (#309)
| Benchmark | Codellama 70b Instruct | GPT-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 1.1% |
| LMArena Math | — | 1269 |
| MATH Level 5 | — | 23% |
| GSM8K | — | 92% |
Knowledge Not comparable
Codellama 70b Instruct: —, GPT-4: 18.4 (#282)
| Benchmark | Codellama 70b Instruct | GPT-4 |
|---|---|---|
| GPQA Diamond | — | 35.7% |
| LMArena Expert | — | 1211 |
| MMLU | — | 86.4% |
| TriviaQA | — | 84.8% |
Multilingual GPT-4 leads
Codellama 70b Instruct: 24.8 (#288), GPT-4: 40.6 (#215)
| Benchmark | Codellama 70b Instruct | GPT-4 |
|---|---|---|
| LMArena Non-English | 992 | 1246 |
| LMArena Chinese | — | 1242 |
| LMArena French | — | 1283 |
| LMArena German | — | 1251 |
| LMArena Japanese | — | 1209 |
| LMArena Korean | — | 1184 |
| LMArena Russian | — | 1251 |
| LMArena Spanish | — | 1261 |
Instruction Following GPT-4 leads
Codellama 70b Instruct: 51.9 (#293), GPT-4: 65.3 (#222)
| Benchmark | Codellama 70b Instruct | GPT-4 |
|---|---|---|
| LMArena Instruction Following | 1024 | 1241 |
Long Context Not comparable
Codellama 70b Instruct: —, GPT-4: 37.7 (#212)
| Benchmark | Codellama 70b Instruct | GPT-4 |
|---|---|---|
| LMArena Longer Query | — | 1244 |
Writing & Preference GPT-4 leads
Codellama 70b Instruct: 33.4 (#277), GPT-4: 34.9 (#268)
| Benchmark | Codellama 70b Instruct | GPT-4 |
|---|---|---|
| LMArena Text | 1057 | 1263 |
| LMArena Creative Writing | — | 1244 |
| EQ-Bench Creative Writing | — | 752 |
| LMArena Multi-Turn | — | 1257 |
Frequently asked questions
Is Codellama 70b Instruct better than GPT-4?
Codellama 70b Instruct is the stronger model overall, scoring 33.7 to 29.1 on the Noometry Index.
Is Codellama 70b Instruct or GPT-4 better for coding?
Codellama 70b Instruct scores higher on coding benchmarks: 37.6 versus 31.6 in the Noometry coding category.
How many benchmarks do Codellama 70b Instruct and GPT-4 share?
7 benchmarks have published results for both models. Codellama 70b Instruct has 7 scored results on Noometry and GPT-4 has 38.