whatmodel.sh / StarCoderBase 7B

StarCoderBase 7B

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

StarCoderBase 7B is an open-weights language model from BigCode, released July 2023. Its published size is 7.53B parameters. Licence: bigcode-openrail-m.

Catalogue identity notes

Comparison-only. Maker repo bigcode/starcoderbase-7b @94e20e9: licence tag bigcode-openrail-m. 7,528,589,824 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2023-07: month the maker's Hugging Face repo was created (createdAt 2023-07-26). StarCoderBase (2023) 7B; not StarCoder2.

30.8±0.7#637 of 719 LLMsfrom 16 benchmark results across 4 sources

$ how it compares

The LLMs ranked either side of StarCoderBase 7B, out of 828 with a score, then more from BigCode.

#modelscorewhereRAMreleased
732Qwen-7B-Chat Alibaba31±1local~6.6GB2023-08
733Gemma 3 1B Google31±1local~2.2GB2025-03
734StarCoderBase 7B BigCode31±1local~6GB2023-07
735Llama 2 7B (base) Meta31±1local~5.5GB2023-07
736OPT-66B Meta31±1local~141GB2022-05
more from BigCode
521StarCoder2-15B-Instruct-v0.1 BigCode46±1local~11GB2024-04
639StarCoder2 15B BigCode39±1local~11GB2024-02
698StarCoder2 7B BigCode35±1local~6.1GB2024-02

see all LLMs · find the ones that fit your machine

$ what it's good at

coding▁▁▁▃▆██▇▆▄▃▁ #435 of 458 · 95% of its score · 9 tests

Beats Qwen3.5 2B, Molmo 7B-D (0924), DeepSeek R1 Distill Qwen 1.5B here, though they rank higher overall. Trails Granite 3.0 1B-A400M Instruct, Llama 3.2 1B Instruct, SmolLM2 1.7B Instruct here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
HumanEval +124.4% ✎▁·▁▁▃▄▆▅█▃▁▁ beats 8/99easy · avg 79%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 42%sharp · at its level · several runs
HumanEval+21.3% ✎▁·▁▁▁▃▄▄▄█▃▁ beats 10/99easy · avg 74%▇▇▇▇▇▇▇▇ 21%sharp
MBPP +134.8% ✎█·▅▃▃▃▅▃▃▅·▅ beats 3/15easy · avg 82%▇▇▇▇ 9.2%at its level · several runs
HumanEvalSynthesize (HumanEvalPack, 6-language average)26.7% ✎▁·▂▁▃▆▅█▄▄▁▂ beats 4/35medium · avg 64%▇▇ 6.1%at its level
CRUXEval-I +129.7% ✎▁··▁▅▂█▇▁··▁ beats 3/22medium · avg 59%▇▇ 4.4%at its level · several runs
CRUXEval-O +132.2% ✎▁·▂▃▁█▁▂···▁ beats 3/22medium · avg 55%▇▇ 3.8%blunt · at its level · several runs
DS-1000 lab19.1% ✎▄▄·▄██▄█·█·▄ beats 2/12medium · avg 52%▇ 3.6%typical
HumanEvalExplain (HumanEvalPack, 6-language average)14.8% ✎▄▇█▃▄▇▄▁▇▃·▁ beats 8/35medium · avg 49%▇ 2.9%typical
1 more test
testresultvs the others in its rundifficultycountswhy
HumanEvalFix (HumanEvalPack, 6-language average)12.5% ✎▁·▁▁▄██▄█▁▅▅ beats 9/35medium · avg 56%▇ 2.2%sharp
math▃▂▃▅▆▅▄█▄▁▁▁ #397 of 398 · 4.6% of its score · 2 tests

Trails Qwen2 0.5B Instruct, Granite 4.0 350M, Granite 4.0 H 350M here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
GSM8K lab +114.9% ✎▄▄▄·▄█·██·█▄ beats 2/12easy · avg 75%▇ 3.6%sharp · at its level · several runs
MATH2.4% ✎▃·▃····▃···█ beats 0/5medium · avg 40%▇ 1.0%noisy

$ where to run it

Ollama · recommended

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run StarCoderBase 7B · needs ~4.3 GB · from ollama.com
    chat in the terminal
    ollama run starcoder:7b

Other ways to run it

  1. Run it as published · needs ~30.11 GB · from huggingface.co/bigcode (official)
    uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='bigcode/starcoderbase-7b', torch_dtype='auto', device_map='auto')('def print_hello_world():', max_new_tokens=40)[0]['generated_text'])"

Official weights and documentation: huggingface.co/bigcode ↗

$ how the score was worked out · EvalPlus / EvalPerf published… 52%, Granite Code Models: A Family… 23%, StarCoder 2 and The Stack v2:… 22% and 1 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
EvalPlus / EvalPerf published evaluation runs2▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 52%Apache-2.0
Granite Code Models: A Family of Open Foundation Models for Code Intelligence — authors' own runs8▇▇▇▇▇▇▇ 23%CC BY 4.0
StarCoder 2 and The Stack v2: The Next Generation — authors' own runs · lab-reported4▇▇▇▇▇▇▇ 22%CC BY-SA 4.0
CRUXEval published evaluation results (authors' generations and scores)2▇ 3%MIT
prior · pull toward the middle▇ 0%
weighted average= 30.8

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json