whatmodel.sh / ERNIE 4.5 300B-A47B

ERNIE 4.5 300B-A47B

+ contribute

ERNIE 4.5 300B-A47B is an open-weights language model from Baidu, released June 2025. Its published size is 300B parameters (47B active per token) and a 131k-token context window. Licence: apache-2.0.

59.9±1.3#196 of 719 LLMsfrom 18 benchmark results across 1 source

$ how it compares

The LLMs ranked either side of ERNIE 4.5 300B-A47B, out of 828 with a score, then more from Baidu.

#	model	score	where	RAM	released
233	Falcon-H1R 7B TII	60±1	local	~6GB	2026-01
234	Gemini 2.0 Flash (001) Google	60±1	comparison only		2025-02
235	ERNIE 4.5 300B-A47B Baidu	60±1	local	~97GB	2025-06
236	Mistral Medium 3 (2505) Mistral	60±1	comparison only		2025-05
237	Qwen3 235B-A22B (Apr 2025, hybrid) Alibaba	60±1	api $1.23/M +local	~76GB	2025-04
	more from Baidu
378	ERNIE 4.5 21B-A3B Baidu	52±1	comparison only		2025-06

see all LLMs · find the ones that fit your machine

$ what it's good at

coding······█····· #1 of 1 · 15% of its score · 3 tests

test	result	vs the others in its run	difficulty	counts	why
HumanEval+ lab	92.1% ✎	▁·▁▁▂▃▄▄█▅▄▃ beats 138/145	easy · avg 83%	▇▇▇ 6.4%	sharp · maxed out · small run
LiveCodeBench (version unspecified) lab	38.8% ✎	▁···▁▁▃▅▅█▄▅ beats 7/37	hard · avg 28%	▇▇ 5.8%	at its level · small run
MBPP+ lab	81.0% ✎	▁··▁▁▃▄█▆▄▃▁ beats 115/120	easy · avg 70%	▇ 3.2%	small run

chat & writing······█····· #1 of 1 * · 2.7% of its score · 1 test

test	result	vs the others in its run	difficulty	counts	why
IFEval (aggregation unspecified) lab	88.0% ✎	▁▁▃▁▂▄▄▆█▃▂▁ beats 95/141	easy · avg 78%	▇ 2.7%	small run

math······█····· #1 of 1 · 33% of its score · 4 tests

test	result	vs the others in its run	difficulty	counts	why
AIME 2025 lab	35.1% ✎	▁·▁▁▁▂▅▆█▇▅▁ beats 22/133	hard · avg 21%	▇▇▇▇ 11%	sharp · at its level · small run
AIME 2024 lab	54.8% ✎	▁▁▁▂▁▃▁▅███▃ beats 25/85	hard · avg 31%	▇▇▇▇ 10%	sharp · at its level · small run
GSM8K lab	96.6% ✎	▂▂▂▅▅▆▅▇█▇▄▃ beats 145/146	very easy · avg 90%	▇▇▇ 7.0%	sharp · maxed out · small run
MATH-500 lab	96.4% ✎	▁▃▂▃▅▇▅▇█▅▂▁ beats 61/84	very easy · avg 87%	▇▇ 4.5%	maxed out · small run

other tests10 not in a task · 48% of its score

test	result	vs the others in its run	difficulty	counts	why
BIG-Bench Hard lab	94.3% ✎	▁▁▁▁▂▄▅▅█▃▄▃ beats 129/129	easy · avg 79%	▇▇▇ 8.1%	sharp · small run
MMLU lab	86.5% ✎	▁▁▁▁▁▂▃▅█▃▆▃ beats 166/183	easy · avg 74%	▇▇▇ 6.4%	small run
SimpleQA lab	45.9% ✎	▄▄▄▆█▆███·▃█ beats 28/32	hard · avg 16%	▇▇ 6.2%	at its level · small run
MMLU-Pro lab	78.4% ✎	▁▁·▁▁▁▃▂▄▅█▁ beats 117/188	medium · avg 60%	▇▇ 5.7%	small run
C-Eval lab	90.6% ✎	▁···▁▁▂▁▂▂▃█ beats 26/33	easy · avg 79%	▇▇ 5.4%	small run
CMMLU lab	90.2% ✎	▅·▅····█▃▅█▅ beats 14/14	easy · avg 79%	▇▇ 4.4%	small run
ChineseSimpleQA lab	77.1% ✎	▄··▄▄▄··▄··█ beats 4/6	medium · avg 52%	▇ 3.7%	at its level · small run
SysBench (aggregation unspecified) lab	69.8% ✎	█··█····█··█ beats 1/3	medium · avg 52%	▇ 3.5%	at its level · small run

2 more tests

test	result	vs the others in its run	difficulty	counts	why
Multi-IF (aggregation unspecified) lab	76.6% ✎	▁▁▁▁▁▁▃▄█▄▁▂ beats 23/41	medium · avg 62%	▇ 2.6%	blunt · at its level · small run
MuSR lab	69.9% ✎	█·██······██ beats 4/4	medium · avg 58%	▇ 2.1%	blunt · at its level · small run

$ 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 | sh
Windows
irm https://llama.app/install.ps1 | iex
other ways to install ↗
Run ERNIE 4.5 300B-A47B · needs ~180.5 GB · from huggingface.co/unsloth
chat in the terminal
llama cli -hf unsloth/ERNIE-4.5-300B-A47B-PT-GGUF:UD-Q4_K_XL
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf unsloth/ERNIE-4.5-300B-A47B-PT-GGUF:UD-Q4_K_XL

LM Studio

version

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run ERNIE 4.5 300B-A47B · needs ~180.5 GB · from huggingface.co/unsloth
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/unsloth/ERNIE-4.5-300B-A47B-PT-GGUF@UD-Q4_K_XL -y
Then run lms ls to see its name, and lms chat followed by that name

MLX (Mac)

Install MLX (Mac) · Apple silicon Macs only, text only.
with uv
uv tool install mlx-lm
or
pip install mlx-lm
Run ERNIE 4.5 300B-A47B · needs ~168.5 GB, Mac only · from huggingface.co/mlx-community
chat in the terminal
mlx_lm.chat --model mlx-community/ERNIE-4.5-300B-A47B-PT-4bit
Or start an OpenAI-compatible server at http://localhost:8080/v1
mlx_lm.server --model mlx-community/ERNIE-4.5-300B-A47B-PT-4bit

Official weights and documentation: huggingface.co/baidu ↗

$ how the score was worked out · ERNIE 4.5 post-trained 300B-A… 99%

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

source	results	share of the score	licence
ERNIE 4.5 post-trained 300B-A47B benchmark chart · lab-reported	18	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 99%	Apache-2.0
prior · pull toward the middle		▇ 1%
weighted average		= 59.9

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit 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 2026-10-11 02:20 UTC · sources and licences · json