Models · Compare
GPT-3.5 Turbo vs Llama 2 70B Chat
Rows highlighted in warm gray are where the models differ. Numbers carry their as-of date and primary source.
Specs
| Field | A: GPT-3.5 Turbo | B: Llama 2 70B Chat |
|---|---|---|
| Released | — | 2023-07-18 |
| Developer | OpenAI | Meta |
| Openness | Proprietary | Source-available |
| License | Proprietary | Llama 2 Community License |
| OSI-approved | no | no |
| Data released | no | no |
| Training code | no | no |
| Architecture | unknown | dense |
| Total params | — | 70B |
| Active params | — | — |
| Experts | — | — |
| Context window | — | — |
| Attention | unknown | gqa |
| Position enc. | unknown | rope |
| Pretraining tokens | — | — |
| Post-training | rlhf | sft, rlhf |
| Training hardware | — | A100 |
| $/M input | $0.50 | — |
| $/M output | $1.50 | — |
| Output tok/sec | 91.9 | — |
Benchmarks
Missing scores render as not reported; never inferred. Bold highlights the leader per benchmark.
General reasoning
| MMLU-Pro | 46.2 2026-05-21 | — |
| GPQA-Diamond | 29.7 2026-05-21 | — |
Code
Math
| MATH | 44.1 2026-05-21 | — |
Context · A
Launched alongside ChatGPT on November 30, 2022 as the chat-optimized successor to text-davinci-003 at roughly one-tenth the cost. The 0125 revision settled on a 16K context window and became the default low-cost endpoint through the GPT-4 era.
Context · B
The first Meta-blessed commercial-use open-weights release. Llama 2's community license permitted commercial use under a 700M-MAU cap, opening the door to a year of derivatives that defined what open-weights deployment looked like.