Open-source releaseCritical global significanceConfirmed confidence
DeepSeek starts with an open-weight code model
DeepSeek releases code models from 1.3B to 33B parameters, making open weights and code part of its product strategy from the outset.
What happened
Event details
DeepSeek Coder was trained from scratch on two trillion tokens and released across sizes from 1.3B to 33B, with a 16K context window and fill-in-the-middle code completion. The repository code used the MIT license, while the weights used the DeepSeek Model License with commercial use allowed. This was not a consumer chatbot launch; it was a developer-first entry designed to earn credibility in the open-model community.
Assessment
Why it matters
DeepSeek's first release established the open-weight playbook that later made V3 and R1 spread quickly among developers.
86/100Global significance score. Regional effects are recorded only when the evidence supports a meaningful difference.