4 AI models similar to CodeLlama 70B by Meta, ranked by performance.
| # | Model | Provider | Score | Rating | Context | Input $/1M | Compare |
|---|---|---|---|---|---|---|---|
| 1 | Qwen 2.5 Coder 32B Alibaba's specialized code model that matches GPT-4o on many coding benchmarks while being open-weight. | Alibaba | 88.0 | ★ 4.5 | 128K | — | vs CodeLlama 70B |
| 2 | Codestral Mistral AI's dedicated code generation model optimized for code completion, generation, and debugging across 80+ programming languages. | Mistral AI | 86.2 | ★ 4.6 | 32K | $0.20 | vs CodeLlama 70B |
| 3 | DeepSeek Coder V2 DeepSeek's specialized code model with strong performance on coding benchmarks at competitive pricing. | DeepSeek | 84.5 | ★ 4.5 | 128K | $0.14 | vs CodeLlama 70B |
| 4 | StarCoder 2 15B Open-source code model trained on The Stack v2, supporting 600+ programming languages for broad code generation tasks. | BigCode | 72.0 | ★ 4.0 | 16K | — | vs CodeLlama 70B |