StarCoder
The original BigCode open-science coder — 15.5B trained on a trillion tokens of permissively licensed code.
About
StarCoder (May 2023) is the original BigCode open-science code model — 15.5B parameters trained on a trillion tokens of permissively licensed source code from The Stack v1.2, covering 80+ programming languages. At release it was the best open-access code LLM available, and its fill-in-the-middle training made it a workhorse for IDE completion. Released under the BigCode OpenRAIL-M license, which carries use-case restrictions (for example, against generating malicious code) — a responsible-AI license, not an open-source one. Entirely superseded by StarCoder2 and later coders, but historically the model that proved open, transparently-sourced code LLMs could compete.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Team-reported: best open-access code LLM at release; strong HumanEval and MultiPL-E scores for its generation.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: BusinessWire release announcement + StarCoder2 paper context read this session: May 4 2023, 15.5B, 1T tokens from The Stack v1.2 (80+ languages), 8K context, BigCode OpenRAIL-M license with use-case restrictions.; Team-reported: best open-access code LLM at release. Superseded by StarCoder2 — documented as historical entry..