The largest model in the DeepSeek Coder series was the first open-source model to beat GPT-3.5 Turbo on competitive programming benchmarks, scoring 27.8% pass@1 on LeetCode Contest problems. It outperforms CodeLlama-34B by 7.9% on HumanEval Python despite sharing a similar parameter count, trained on 2 trillion tokens with an additional 200B-token context extension phase.
DeepSeek Coder 33B Instruct scores 79.3% on HumanEval and 50.3% on HumanEval multilingual (8-language average), outperforming Code Llama 70B Instruct (67.8% HumanEval) on the same sheet despite being half the size. It was the first open-source model to beat GPT-3.5 Turbo on LeetCode benchmarks at 27.8% pass@1. Standard MMLU is not published for this code-focused model.
The cost of running DeepSeek Coder 33B with Telnyx Inference is $0.0003 per 1,000 tokens. Generating code for 1,000,000 programming tasks at 1,000 tokens each would cost $300, half the cost of Code Llama 70B ($600) while outperforming it on HumanEval (79.3% vs 67.8%).
Discover the power and diversity of large language models available with Telnyx. Explore the options below to find the perfect model for your project.
| Organization | Model Name | Tasks | Languages Supported | Context Length | Parameters | Model Tier | License |
|---|---|---|---|---|---|---|---|
| No data available at this time, please try again later. |
Powered by our own GPU infrastructure, select a large language model, add a prompt, and chat away. For unlimited chats, sign up for a free account on our Mission Control Portal here.
Check out our helpful tools to help get you started.
DeepSeek Coder 33B Instruct is the instruction-tuned variant of the DeepSeek Coder family, fine-tuned for following natural language instructions to generate code. The base model and instruct variant serve different use cases.
DeepSeek Coder 33B Instruct is the strongest model in the original DeepSeek Coder series for code generation tasks. For newer alternatives, DeepSeek V2 and V3 offer improved performance, available on Hugging Face.
DeepSeek Coder 33B can be accessed through hosted inference providers or deployed locally using the Hugging Face Transformers library. Telnyx offers API access for production code generation workloads.
DeepSeek Coder models are implemented in Python using PyTorch, but they generate code in multiple programming languages including Python, Java, C++, JavaScript, and many others.
DeepSeek Coder 33B Instruct performs well on code generation benchmarks, competing with Code Llama 34B at the time of release. It handles code completion, generation, and infilling across multiple languages through hosted inference platforms.
DeepSeek Coder 33B is released under a permissive license that allows free commercial use. Weights are available on Hugging Face for self-hosting, and hosted inference is available through various providers.
