whatmodel.sh / Phi-3-medium-4k-instruct

Phi-3-medium-4k-instruct

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

Phi-3-medium-4k-instruct is an open-weights language model from Microsoft, released May 2024. Its published size is 14B parameters and a 4k-token context window. Licence: mit.

Catalogue identity notes

Arena entry 'phi-3-medium-4k-instruct' (text). LMArena's leaderboard table (arena-leaderboard space @6f2eeeb) labels 'phi-3-medium-4k-instruct' as 'Phi-3-Medium-4k-Instruct' (Microsoft, MIT), linking https://huggingface.co/microsoft/Phi-3-medium-4k-instruct. Maker repo microsoft/Phi-3-medium-4k-instruct @48d87cd: licence tag mit. 13,960,238,080 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2024-05: card: 'The model weight is released on May 21, 2024'.

44.9±1.1#445 of 719 LLMsfrom 7 benchmark results across 7 sources

$ how it compares

The LLMs ranked either side of Phi-3-medium-4k-instruct, out of 828 with a score, then more from Microsoft.

#modelscorewhereRAMreleased
531Claude 2.1 Anthropic45±1comparison only2023-11
532Zephyr ORPO 141B-A35B v0.1 Hugging Face45±2local~56GB2024-04
533Phi-3-medium-4k-instruct Microsoft45±1local~9.8GB2024-05
534Qwen2-Math 1.5B Instruct Alibaba45±2local~2.4GB2024-08
535Mistral Small (2402) Mistral45±1comparison only2024-02
more from Microsoft
252Phi-4-reasoning-plus 14B Microsoft59±1local~10GB2025-04
261Phi-4-reasoning 14B Microsoft58±1api $0.22/M +local~10GB2025-04
365Phi-4 14B Microsoft53±1api $0.087/M +local~10GB2024-12

see all LLMs · find the ones that fit your machine

$ what it's good at

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #304 of 393 · 45% of its score · 3 tests

Beats Gemma 4 E4B, OpenThinker3 7B, Gemma 4 E2B here, though they rank higher overall. Trails LFM2 350M, Gemma 3 1B, Flan-T5 XXL here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Arena-Hard (v0.1, GPT-4-1106 judge)33.4% ✎▂▁▁▅█▅▄▃▃▅▃▁ beats 29/60medium · avg 52%▇▇▇▇▇▇▇▇▇▇ 24%sharp · at its level
Polyrating rating (bias-aware re-fit of Chatbot Arena votes, mid-2024)1126.0 ✎▂▁▂▁▃▇█▆▇▄▂▂ beats 58/95avg 1201▇▇▇▇ 10%blunt
LMArena (text)1197.9 ✎▁▁▁▂▃▃▃▆▆▇█▃ beats 68/379avg 1268▇▇▇▇ 10%blunt
other tests4 not in a task · 55% of its score
testresultvs the others in its rundifficultycountswhy
LiveBench 2024-07-2631.2% ✎▁▁▁▂▂▃▂█▆▃▅▃ beats 30/61medium · avg 41%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 39%at its level
EvalPerf correctness (118 tasks, 20240328)68.4% ✎▂·▂▂▂▆▅▆▆▆▃█ beats 5/29easy · avg 76%▇▇▇ 7.7%at its level
Polish EQ-Bench v2 (SpeakLeash)57.1% ✎▁··▁▁▂▂▆▅▆█▆ beats 34/70medium · avg 64%▇▇ 6.0%at its level
Open PT-LLM Leaderboard average (9 Portuguese tasks)72.3% ✎▃▁▁▁▁▁▁▁▂█▅▁ beats 194/276medium · avg 73%▇ 2.3%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 Phi-3-medium-4k-instruct · needs ~8.6 GB · from huggingface.co/bartowski
    chat in the terminal
    llama cli -hf bartowski/Phi-3-medium-4k-instruct-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf bartowski/Phi-3-medium-4k-instruct-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 Phi-3-medium-4k-instruct · needs ~8.6 GB · from huggingface.co/bartowski

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/bartowski/Phi-3-medium-4k-instruct-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run Phi-3-medium-4k-instruct · needs ~8.9 GB · from ollama.com
    chat in the terminal
    ollama run phi3:14b-medium-4k-instruct-q4_K_M

MLX (Mac)

version

  1. Install MLX (Mac) · Apple silicon Macs only, text only.
    with uv
    uv tool install mlx-lm
    or
    pip install mlx-lm
  2. Run Phi-3-medium-4k-instruct · needs ~7.9 GB, Mac only · from huggingface.co/mlx-community
    chat in the terminal
    mlx_lm.chat --model mlx-community/Phi-3-medium-4k-instruct-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model mlx-community/Phi-3-medium-4k-instruct-4bit

Official weights and documentation: huggingface.co/microsoft ↗

$ how the score was worked out · LiveBench model judgments (HF… 39%, Arena-Hard-Auto v0.1 leaderbo… 24%, Polyrating: A Cost-Effective… 10% and 4 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
LiveBench model judgments (HF livebench/model_judgment, 2024-10-22 revision) with the LiveBench question datasets1▇▇▇▇▇▇▇▇▇▇▇▇ 39%Apache-2.0 (stated in the LiveBench datasheet)
Arena-Hard-Auto v0.1 leaderboard (repository README, pre-v2)1▇▇▇▇▇▇▇ 24%Apache-2.0
Polyrating: A Cost-Effective and Bias-Aware Rating System for LLM Evaluation (arXiv:2409.00696v3)1▇▇▇ 10%CC-BY-4.0
LMArena leaderboard dataset1▇▇▇ 10%CC BY 4.0
EvalPlus / EvalPerf published evaluation runs1▇▇ 8%Apache-2.0
SpeakLeash Polish EQ-Bench leaderboard1▇▇ 6%Apache-2.0
Open Portuguese LLM Leaderboard (archived)1▇ 2%Apache-2.0
prior · pull toward the middle▇ 1%
weighted average= 44.9

$ 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