whatmodel.sh / Llama 2 7B (base)

Llama 2 7B (base)

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

Llama 2 7B (base) is an open-weights language model from Meta, released July 2023. Its published size is 6.74B parameters and a 4k-token context window. Licence: llama2.

Catalogue identity notes

Pretrained (not chat) Llama 2 checkpoint; the chat models are separate identities (llama-2-*-chat). Ollama 'text' tags are the pretrained model.

30.6±0.5#638 of 719 LLMsfrom 71 benchmark results across 11 sources

$ how it compares

The LLMs ranked either side of Llama 2 7B (base), out of 828 with a score, then more from Meta.

#modelscorewhereRAMreleased
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
737LFM2 350M Liquid AI30±2local~1.7GB2025-07
more from Meta
24Muse Spark 1.3 Meta77±1api $2.00/M2026-09
40Muse Spark 1.2 Meta75±1api $2.00/M2026-08
43Muse Spark 1.1 Meta75±1api $2.00/M2026-07

see all LLMs · find the ones that fit your machine

$ what it's good at

coding▁▁▃▃▆██▇▆▄▃▁ #450 of 458 · 10% of its score · 3 tests

Beats Molmo 7B-D (0924), DeepSeek R1 Distill Qwen 1.5B, Qwen3 0.6B here, though they rank higher overall. Trails Phi-1 (1.3B), OpenAssistant SFT-4 Pythia 12B, Falcon3 1B Instruct here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
HumanEval +211.6% ✎█···█·███··█ beats 0/5easy · avg 66%▇▇ 5.3%sharp · several runs
MBPP +226.1% ✎▄··▄·▄··█··▄ beats 0/5easy · avg 71%▇ 2.9%at its level · several runs
Tulu 2: CodexEval pass@1026.8% ✎▃▃▃·▆▃▃█▃▂▂▅ beats 1/25very easy · avg 86%▇ 1.7%sharp · at its level
math▃▂▃▅▆▅▄█▄▁▁▁ #395 of 398 · 5.5% of its score · 2 tests

Beats DeepSeek-Coder 1.3B Base, StarCoderBase 7B here, though they rank higher overall. 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 +312.0% ✎▆▂▅█▅▅▃▆···▂ beats 2/25easy · avg 82%▇▇ 4.1%sharp · at its level · several runs
MATH +13.9% ✎█·█·█·██···█ beats 0/5hard · avg 30%▇ 1.4%too hard for it · several runs
other tests22 not in a task · 84% of its score
testresultvs the others in its rundifficultycountswhy
ARC Challenge +546.2% ✎▁·▁··▁▁▇▂█▆▃ beats 21/26easy · avg 85%▇▇▇▇▇▇▇▇▇▇ 26%sharp · at its level · several runs
WinoGrande +469.6% ✎▁····▁··▁▂█▄ beats 22/26easy · avg 89%▇▇▇▇ 10%at its level · several runs
HellaSwag +475.9% ✎▃···▁▇▅█▇▃▁▁ beats 19/26very easy · avg 92%▇▇▇ 6.3%at its level · several runs
BoolQ +277.7% ✎▂▃▂▅▆▂▅▆█▃▂▂ beats 24/26easy · avg 91%▇▇ 5.7%at its level · several runs
ARC Easy +474.5% ✎▁▁·▁▁▄▅▄█▅▃▃ beats 24/26easy · avg 88%▇▇ 5.5%at its level · several runs
Natural Questions (closed book) +227.1% ✎▁·▁·▁▂▁▂·▅▂█ beats 23/28medium · avg 58%▇▇ 4.7%several runs
MMLU +341.8% ✎▁▁▃▃▄█▄▁█··▁ beats 3/25easy · avg 82%▇▇ 4.6%at its level · several runs
BIG-Bench Hard +139.3% ✎▁·····▃█▂▅·▁ beats 2/25easy · avg 77%▇▇ 4.2%sharp · several runs
14 more tests
testresultvs the others in its rundifficultycountswhy
OpenBookQA +231.4% ✎█·▁▁▃·▁▃▅▇▃▄ beats 19/26easy · avg 84%▇▇ 4.1%at its level · several runs
PIQA +477.7% ✎▁·▁··▁▁▃█▅█▂ beats 17/26easy · avg 88%▇ 3.3%blunt · at its level · several runs
TriviaQA +29.2% ✎▁··▁█▃▃▁▄▅▄█ beats 4/28very easy · avg 86%▇ 2.8%at its level · several runs
SciQ +193.6% ✎▁·▁·▁·▃▅▃█▄▁ beats 17/26very easy · avg 98%▇ 2.4%several runs
Social IQA +143.5% ✎▁···▁▃█▃▁▁·▁ beats 21/26medium · avg 63%▇ 1.5%blunt · several runs
LAMBADA (OpenAI version), last-word accuracy73.5% ✎▁··▂▁▁▁▂▆█▅▃ beats 21/26very easy · avg 89%▇ 0.7%blunt · at its level
C-Eval32.4% ✎▁···▃▁▂▁▂▂▃█ beats 1/33medium · avg 62%▇ 0.4%noisy
Tulu 2: TyDiQA gold passage, 1-shot F151.2% ✎▂··▁▃▂▂▂█▅▁▂ beats 12/25medium · avg 61%▇ 0.4%blunt · at its level
Open PT-LLM Leaderboard average (9 Portuguese tasks)48.9% ✎▃▁▁▁▁▁▁▃▂█▅▁ beats 65/276medium · avg 73%▇ 0.3%noisy
TruthfulQA MC2 +139.0% ✎▅▃█▆▁▂▅···▁▁ beats 7/31medium · avg 64%▇ 0.3%blunt · at its level · several runs
EuroEval Life in the UK (English knowledge MC), test split, 5-shot, accuracy52.8% ✎▁▁▁▁▁▁▃▂▄█▄▃ beats 21/146easy · avg 83%▇ 0.3%at its level · noisy
EuroEval HellaSwag (English), test split, 5-shot generative MC, accuracy31.3% ✎▃·▁▁▃▁▂▅█▇▄▁ beats 13/100medium · avg 68%▇ 0.2%noisy
HaluEval QA hallucination detection (accuracy)45.7% ✎█▁▁·▁▁·▁▁▁▁▁ beats 5/27medium · avg 70%▇ 0.2%sharp · too hard for it · noisy
COPA82.6% ✎▅·▃·▃··█▃··▃ beats 4/8easy · avg 92%▇ 0.1%at its level · noisy

$ where to run it

llama.cpp · recommended

version

  1. Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
    macOS / Linux
    curl -LsSf https://llama.app/install.sh | sh
    Windows
    irm https://llama.app/install.ps1 | iex

    other ways to install ↗

  2. Run Llama 2 7B (base) · needs ~4.1 GB · from huggingface.co/TheBloke
    chat in the terminal
    llama cli -hf TheBloke/Llama-2-7B-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf TheBloke/Llama-2-7B-GGUF:Q4_K_M

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run Llama 2 7B (base) · needs ~4.1 GB · from huggingface.co/TheBloke

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/TheBloke/Llama-2-7B-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run Llama 2 7B (base) · needs ~3.8 GB · from ollama.com
    chat in the terminal
    ollama run llama2:7b-text

Weights and documentation: huggingface.co/meta-llama ↗

$ how the score was worked out · Stability AI — original Stabl… 29%, Mixtral of Experts — base-mod… 17%, OLMo: Accelerating the Scienc… 14% and 8 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
Stability AI — original StableLM zero-shot evaluation outputs11▇▇▇▇▇▇▇▇▇ 29%Apache-2.0
Mixtral of Experts — base-model evaluations12▇▇▇▇▇ 17%CC BY 4.0
OLMo: Accelerating the Science of Language Models — authors' own runs8▇▇▇▇ 14%CC BY 4.0
Textbooks Are All You Need II: phi-1.5 technical report — authors' own runs8▇▇▇ 11%CC BY 4.0
Mistral 7B — authors' own runs12▇▇▇ 9%CC BY 4.0
MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies (Hu et al.)7▇▇▇ 9%CC BY 4.0
Tulu 2 — base and instruction-tuned evaluations5▇▇ 6%CC BY 4.0
Hallucinations Leaderboard results (HF dataset hallucinations-leaderboard/results)4▇ 3%Apache-2.0
EuroEval leaderboards (results collection), English generative leaderboard2▇ 1%MIT
Open Portuguese LLM Leaderboard (archived)1▇ 0%Apache-2.0
From Test-Taking to Test-Making: Examining LLM Authoring of Commonsense Assessment Items (arXiv:2410.14897v1)1▇ 0%CC-BY-4.0
prior · pull toward the middle▇ 0%
weighted average= 30.6

$ 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