Model
Model explorer

StarCoder (15.5B)

OPEN
BigCode · StarCoder family · released May 4, 2023

Pivotal 15.5B open code LLM trained on The Stack across 80+ languages; matched Codex-cushman on HumanEval/MBPP.

ReasoningCodingVisionFunction callingTool useAgentic
210.7
Elo · rank #361
Parameters
15.5B
Active params
15.5B (dense)
Context
8K tokens
Architecture
GPT-2-style decoder with multi-query attention (GPTBigCode) + fill-in-the-middle, 8K context
License
BigCode OpenRAIL-M
Languages
80+
API price (in/out)
No hosted API
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
DS-1000Coding26.0%#2
best: StarCoder2 15B · 33.8%
FIN-benchKnowledge45.5%#5
best: Poro 34B · 66.3%
HumanEval FIMCoding62.0%#6
best: Codestral 22B · 91.6%
HumanEval+Coding29.3%#35
best: Mistral Small 3.2 (24B) · 92.9%
HumanEvalCoding33.6%#145
best: Claude Opus 4.5 · 99.4%
MBPP+Coding46.1%#37
best: Llama 3.1 405B · 88.6%
MBPPCoding52.7%#68
best: Llama-3.3-Nemotron-Super-49B v1 (Reasoning On) · 91.3%
Run it locally
VRAM @ Q4
9.7 GB
VRAM @ FP16
31.4 GB
Fits on (Q4)
RTX 3060 12GBRTX 4070 Ti 16GBRTX 3090 24GBRTX 4090 24GBRTX 5090 32GBM4 Pro 48GBM3 Max 128GBM3 Ultra 512GBA100 80GBH100 80GBH200 141GBB200 192GB
Throughput data unavailable.
Quantizations
GGUF · GPTQ
StarCoder family
Elo progression across releases
StarCoder (15.5B)May 2023210.7
Resources
HFbigcode/starcoderGHbigcode-project/starcoderDOCSbigcode-project.org/docs/about/models
API price weights · each benchmark row carries its own source badge (see methodology)