Beta version: *Information might not be fully accurate. Please report any discrepancies.

BigCodeVerifiedOpen Weights6 benchmarks

StarCoder2-15B

Released 2024-02-2815B Architecture

Training: 2024-02

Verified Official Model Card

Latest Data

2026-02-18

Context Window

16k

tokens

Input Cost

Free

per 1M tokens

Output Cost

Free

per 1M tokens

Parameters

15B

model footprint

Benchmark Provenance

Performance Analysis // Verified Benchmarks

MMLU (5-shot)Knowledge

45.2*/ 100

Verified

Last Verified: 2026-02-16Artificial Analysis (Independent)

Massive Multitask Language Understanding covers 57 subjects across STEM, the humanities, social sciences, and more.

HumanEvalCoding

72.6/ 100

Verified

Last Verified: 2024-02-28BigCode Project

Functional correctness of synthesized programs from docstrings.

BigCodeBenchCoding

28.7*/ 100

Verified

Last Verified: 2026-02-16Artificial Analysis (Independent)

Next-generation HumanEval with more diverse library calls and complex tasks.

LMArena ELOReal-world

1105/ 1700

Verified

Last Verified: 2026-02-18Chatbot Arena Leaderboard

Chatbot Arena ELO score. Crowd-sourced human preference ranking.

LiveCodeBench v6Coding

24.5*/ 100

Verified

Last Verified: 2026-02-16Artificial Analysis (Independent)

Contamination-free coding benchmark using recent problems.

GPQA DiamondSTEM

28.5*/ 100

Verified

Last Verified: 2026-02-16Artificial Analysis (Independent)

Graduate-Level Google-Proof Q&A Benchmark.