whatmodel.sh / Llama 2 7B Chat
Llama 2 7B Chat
↗ share
+ contribute
Llama 2 7B Chat 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
Comparison model from the Epoch AI internal-evaluation CSVs (Epoch release date 2023-07-18). Chat checkpoint Llama-2-7b-chat (not the base model). 6,738,417,664 parameters (safetensors metadata of the -hf repo).
$ how it compares
The LLMs ranked either side of Llama 2 7B Chat, out of 828 with a score, then more from Meta.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 739 | Llama 3.2 1B Instruct | 30±1 | api $0.020/M +local | ||
| 740 |  Falcon 7B (base) | 30±1 | local | ||
| 741 | Llama 2 7B Chat | 30±1 | local | ||
| 742 |  Baichuan2-7B-Base | 30±1 | local | ||
| 743 | LLaMA 7B | 30±1 | comparison only | ||
| more from Meta | |||||
| 24 | Muse Spark 1.3 | 77±1 | api $2.00/M | ||
| 40 | Muse Spark 1.2 | 75±1 | api $2.00/M | ||
| 43 | Muse Spark 1.1 | 75±1 | api $2.00/M | ||
$ what it's good at
coding▁▃▁▃▆██▇▆▄▃▁ #452 of 458 * · 5.6% of its score · 1 test
Beats Molmo 7B-D (0924), DeepSeek R1 Distill Qwen 1.5B 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.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Tulu 2: CodexEval pass@10 | 24.0% ✎ | ▃▃▃·▆▃▃█▃▂▂▅ beats 0/25 | 5.6% |
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #377 of 393 · 38% of its score · 7 tests
Beats Phi-2 (2.7B), Nous-Hermes-13b, Qwen1.5 4B Chat here, though they rank higher overall. Trails Gemma 3 270M IT, SmolLM2 360M Instruct, Granite 4.0 350M here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| AlpacaEval 2 length-controlled | 5.3% ✎ | ▁▂▃▃▆▆█▆▄▄▁▁ beats 8/122 | 7.9% |
| Polyrating rating (bias-aware re-fit of Chatbot Arena votes, mid-2024) | 1036.0 ✎ | ▂▁▂▁▃▇█▆▇▄▂▂ beats 21/95 | 7.4% |
| LMArena (text) | 1107.9 ✎ | ▁▁▁▃▃▃▃▆▆▇█▃ beats 16/379 | 7.3% |
| MT-Bench (GPT-4 single-answer grading, LMSYS runs) | 6.2700 MT-Bench score (1-10) ✎ | ▆▂▂▆▆▄██▂▄▄▆ beats 14/28 | 6.4% |
| AlpacaEval 1.0 win rate (vs text-davinci-003, GPT-4 judge) | 71.4% ✎ | ▂▂·▆·▇▃▆█▆▄▂ beats 16/45 | 4.4% |
| Arena-Hard (v0.1, GPT-4-1106 judge) | 4.6% ✎ | ▃▁▁▅█▅▄▃▃▅▃▁ beats 2/60 | 2.5% |
| Arabic IFEval prompt-level (Inception) | 0.7% ✎ | ▃·▁▁▂▁▄▄▆▄█▄ beats 0/38 | 1.8% |
math▁▃▃▅▆▅▄█▄▁▁▁ #386 of 398 * · 2.7% of its score · 1 test
Beats Qwen2-VL 2B Instruct, Code Llama 7B (base), Llama 2 13B Chat here, though they rank higher overall. Trails Qwen2 0.5B Instruct, Granite 3.1 1B-A400M Instruct, Gemma 2B (base) here, though they rank lower overall.
other tests16 not in a task · 54% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| LiveBench 2024-07-26 | 11.6% ✎ | ▁▁▁▃▂▃▂█▆▃▅▃ beats 6/61 | 12% |
| Natural Questions (closed book) | 23.7% ✎ | ▁·▁·▁▂▁▂·▅▃█ beats 18/28 | 8.6% |
| WildBench v2 WB-Score (task macro) | 8.2621 WB-Score ✎ | ▁·▁▃·▃█▄▇▆▅▆ beats 2/52 | 8.3% |
| BIG-Bench Hard | 25.6% ✎ | ▃·····▂█▂▅·▁ beats 0/25 | 6.8% |
| MMLU | 46.8% ✎ | ▁▁▃▃▄█▄▁█··▁ beats 7/25 | 2.9% |
| Open Persian LLM Leaderboard v2 average (21 tasks) | 31.7% ✎ | ▃▄▂▃█▄▃▆▃▅▂▂ beats 2/81 | 2.9% |
| EQ-Bench v1 (60 questions, paper run) | 25.4% ✎ | ▂··▄·▂▆▄▂█·▂ beats 2/14 | 2.7% |
| BRIDGE clinical text, zero-shot average (87 tasks) | 16.5% ✎ | ▁▁▁▃▁▂▂▃▄▆█▃ beats 6/97 | 2.3% |
8 more tests
| test | result | vs the others in its run | counts |
|---|---|---|---|
| TriviaQA | 8.8% ✎ | ▁··▁█▃▃▁▄▅▄█ beats 3/28 | 1.3% |
| Tulu 2: TyDiQA gold passage, 1-shot F1 | 22.7% ✎ | ▃··▁▃▂▂▂█▅▁▂ beats 0/25 | 1.2% |
| Open PT-LLM Leaderboard average (9 Portuguese tasks) | 52.2% ✎ | ▃▁▁▁▁▁▁▃▂█▅▁ beats 72/276 | 1.1% |
| EuroEval Life in the UK (English knowledge MC), test split, 5-shot, accuracy | 54.3% ✎ | ▁▁▁▁▁▁▁▃▄█▄▃ beats 25/146 | 1.1% |
| Vectara hallucination leaderboard, 2025 dataset (HHEM-2.3, factual consistency rate) | 88.7% ✎ | ▁▁▂▃▄▄█▆▃▂▂▁ beats 19/140 | 1.0% |
| TruthfulQA MC2 | 45.3% ✎ | ▅▃█▆▁▂▅···▁▁ beats 20/31 | 0.6% |
| HaluEval QA hallucination detection (accuracy) | 52.3% ✎ | █▁▁·▁▃·▁▁▁▁▁ beats 20/27 | 0.5% |
| Chess Puzzles (Epoch) | 0.0% ✎ | █▁▁▁▁▂▁▄▂▂▁▁ beats 8/135 | 0.0% |
$ where to run it
llama.cpp · recommended
version
- 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 | shWindowsirm https://llama.app/install.ps1 | iex - Run Llama 2 7B Chat · needs ~4.1 GB · from huggingface.co/TheBlokechat in the terminal
llama cli -hf TheBloke/Llama-2-7B-Chat-GGUF:Q4_K_M - Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf TheBloke/Llama-2-7B-Chat-GGUF:Q4_K_M
LM Studio
version
- Install LM Studio · Open the app once before using the lms command.
- Run Llama 2 7B Chat · needs ~4.1 GB · from huggingface.co/TheBloke
- Or download it in the terminal
lms get https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGUF@Q4_K_M -y - Then run lms ls to see its name, and lms chat followed by that name
Ollama
version
- Install Ollama
download it (macOS / Windows) ↗
Linuxcurl -fsSL https://ollama.com/install.sh | sh - Run Llama 2 7B Chat · needs ~4.54 GB · from ollama.comchat in the terminal
ollama run llama2:7b-chat
Weights and documentation: huggingface.co/meta-llama ↗
$ how the score was worked out · Tulu 2 — base and instruction… 19%, LiveBench model judgments (HF… 12%, AlpacaEval 1.0 and 2.0 leader… 12% and 14 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
$ improve this page
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