Open-Source Code Generation Models Catalog

AuthorEmmanuel Secretaria

Published Nov 12, 2025

Open-Weight AI Models: These models release their trained weights publicly but often under custom “source-available” licenses rather than standard open-source licenses. Open-Source AI Models: These models are released under OSI-approved open-source licenses (e.g. MIT License, Apache 2.0).

Share

This document lists open and open-weight code-generation models available as of 2025, sorted by size, context window, and license.

Model NameOrganizationSize & ContextLicense / NotesLink
CodeLlama‑7B‑InstructMeta AI7B params, ~16K tokens contextCommunity License (Llama family)https://theorionai.com/blog/top-open-llms
CodeLlama‑13B‑InstructMeta AI13B params, ~16K tokensCommunity License (Llama family)https://theorionai.com/blog/top-open-llms
CodeLlama‑34B‑InstructMeta AI34B params, ~16K tokensCommunity License (Llama family)https://e2enetworks.com/blog/top‑8‑open‑source‑llms‑for‑coding
CodeLlama‑70B‑InstructMeta AI70B params, context maybe <16KCommunity License (Llama family)https://e2enetworks.com/blog/top‑8‑open‑source‑llms‑for‑coding
StarCoder‑15BBigCode~15B paramsApache 2.0 (open code‐model)https://theorionai.com/blog/top‑open‑llms
CodeGen2.5‑7B‑InstructSalesforce AI Research~7B paramsOpen code generation focushttps://github.com/prthm786/awesome-genai-models
DeepSeek‑Coder‑6.7B‑InstructDeepSeek AI~6.7B params, long context (≥64K)Source/weight open (check repo)https://www.reddit.com/r/LangChain/comments/1ite411
Qwen2.5‑Coder‑7B‑InstructAlibaba Cloud7B params, context ~128K tokensOpen‑weight under Apache 2.0https://www.reddit.com/r/ChatGPTCoding/comments/1ite5bb
Qwen3‑CoderAlibaba Mail32B params, new architectureAnnounced July 2025, open‑source modelhttps://www.reuters.com/world/china/alibaba-launches‑open‑source‑ai‑coding‑model‑touted‑its‑most‑advanced‑date‑2025‑07‑23/
CodeGeeX‑13BUnknown / Community~13B paramsOpen model for code generationhttps://www.edenai.co/post/top-free-code-generation-tools-apis-and-open‑source‑models
CodeT5+‑16BSalesforce / Community~16B paramsEncoder‑decoder model for code taskshttps://github.com/prthm786/awesome-genai-models
Granite‑Code‑3B/8B/20B/34BIBM ResearchMultiple sizes: 3B,8B,20B,34BDecoder‐only code model familyhttps://github.com/prthm786/awesome-genai-models
WaveCoder‑Ultra‑6.7BMicrosoft Research~6.7B paramsCode generation & repair specialtyhttps://www.reddit.com/r/LangChain/comments/1ite411
Free‑GPT‑EngineerCommunitySmaller model tuned for project generationMIT licensehttps://www.edenai.co/post/top-free-code-generation-tools-apis-and‑open‑source‑models
DuckargsCommunitySmall model for CLI argument generationOpen licensehttps://www.edenai.co/post/top-free-code-generation-tools-apis-and‑open‑source‑models
CodeBERTMicrosoft Research~0.5‑1B paramsMultilingual code understanding modelhttps://www.edenai.co/post/top-free-code-generation-tools-apis-and‑open‑source‑models
Lambda / LLaMA 3 (Code fine‑tune)Meta AI8B / 70B paramsCode fine‐tuned variant of LLaMA3https://en.wikipedia.org/wiki/Llama_%28language_model%29
BLOOM (multilingual + code)BigScience176B paramsMultilingual, includes code languageshttps://en.wikipedia.org/wiki/BLOOM_%28language_model%29
Gemma / Gemini familyGoogle DeepMindMultiple sizesMultimodal, some code taskshttps://en.wikipedia.org/wiki/Gemma_%28language_model%29
GPT‑OSS‑120B / 20BOpenAI120B / 20B paramsOpen‐weight (released 2025), general tasks including codehttps://timesofindia.indiatimes.com/technology/tech‑news/openai‑launches‑new‑open‑source‑ai‑models-gpt‑oss‑120b-and‑gpt-oss‑20b