GLM-4.7

+ contribute

GLM-4.7 is an open-weights language model from Z.ai, released December 2025. Its published size is 357B parameters (32B active per token) and a 205k-token context window. Licence: mit.

68.7±0.7#98 of 719 LLMsfrom 48 benchmark results across 11 sources

$ how it compares

The LLMs ranked either side of GLM-4.7, out of 828 with a score, then more from Z.ai.

#	model	score	where	RAM	released
105	Qwen3.5 122B-A10B Alibaba	69±1	api $1.10/M +local	~75GB	2026-02
106	Gemini 3.5 Flash-Lite Google	69±1	api $0.85/M		2026-07
107	GLM-4.7 Z.ai	69±1	api $1.00/M +local	~58GB	2025-12
108	GPT-5.5 Instant OpenAI	69±1	comparison only		2026-05
109	GPT-5.4 nano OpenAI	69±1	api $0.46/M		2026-03
	more from Z.ai
33	GLM-5.3 Z.ai	76±1	api $2.15/M +local	~239GB	2026-08
52	GLM-5.3-Flash Z.ai	74±1	api $0.24/M +local	~103GB	2026-08
57	GLM-5.2 Z.ai	74±1	api $2.15/M +local	~121GB	2026-06

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #90 of 300 · 37% of its score · 11 tests

Beats Grok 4.20 (0309 reasoning), GPT-5.1, Grok 4.3 here, though they rank higher overall. Trails Hy3 preview (295B-A21B), Qwen3.6 35B-A3B, MiniMax-M2.7 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
MCP-Atlas (public set) lab +1	52.0% ✎	▄······▄▄▄▄█ beats 0/6	hard · avg 25%	▇▇ 5.9%	at its level · several runs
SWE-bench Verified lab +1	73.8% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 94/131	hard · avg 25%	▇▇ 5.0%	sharp · at its level · small run · several runs
Terminal-Bench 2.0 lab +1	41.0% ✎	▂▂▅▅████▆▂▃▃ beats 14/36	very hard · avg 8.4%	▇▇ 4.0%	at its level · small run · several runs
τ²-bench aggregate lab +1	87.4% ✎	▁▁▁·▂▂▄▃▃▅█▂ beats 34/37	medium · avg 48%	▇▇ 3.8%	small run · several runs
BrowseComp-ZH lab +1	66.6% ✎	▁▁▁▁▁▃·▁▄▅▄█ beats 18/23	hard · avg 22%	▇▇ 3.8%	at its level · small run · several runs
SWE-bench Multilingual lab +1	66.7% ✎	▁··▁▃▂▅▇█▃▃▁ beats 26/60	hard · avg 28%	▇ 3.2%	at its level · small run · several runs
BrowseComp lab +1	52.0% ✎	▂·▄▁▂▃▆▄█▂▃▄ beats 20/42	hard · avg 18%	▇ 3.0%	at its level · small run · several runs
Terminal-Bench Hard lab	33.3% ✎	▂▅▆▂▂▆▂▂█▅▅▂ beats 22/27	very hard · avg 4.5%	▇ 2.7%	sharp · at its level · small run

3 more tests

test	result	vs the others in its run	difficulty	counts	why
Humanity's Last Exam with tools lab +1	42.8% ✎	▂▂▂▁▄▂▃▆█▇▂▆ beats 19/54	hard · avg 16%	▇ 2.2%	at its level · small run · several runs
BrowseComp with context management lab +1	67.5% ✎	▅·▅▃·█▃·█▅·▃ beats 3/14	medium · avg 41%	▇ 2.1%	at its level · small run · several runs
Toolathlon (Tool Decathlon) lab	23.8% ✎	▂▃▄▂▄▄▆▂█▆██ beats 1/27	very hard · avg 11%	▇ 1.0%	at its level · small run

coding▁▁▁▃▆██▇▆▄▃▁ #74 of 458 · 17% of its score · 4 tests

Beats GPT-5.2, GPT-5, Grok 4.20 (0309 reasoning) here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
Arena WebDev (Code Arena)	1434.5 ✎	▁▂▄▃▅█▄▇▆▄▁▁ beats 54/119	avg 981	▇▇▇ 7.3%	sharp
SWE-bench Verified lab +1	73.8% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 94/131	hard · avg 25%	▇▇ 5.0%	sharp · at its level · small run · several runs
SWE-bench Multilingual lab +1	66.7% ✎	▁··▁▃▂▅▇█▃▃▁ beats 26/60	hard · avg 28%	▇ 3.2%	at its level · small run · several runs
LiveCodeBench v6 (window unspecified) lab	84.9% ✎	▂▃▄▄▅██▅▄▇▃▆ beats 51/65	medium · avg 47%	▇ 1.9%	small run

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #62 of 393 * · 4.1% of its score · 1 test

Beats GPT-5.2, Qwen3.8 27B, LongCat 2.0 (1.6T-A48B) here, though they rank higher overall. Trails Gemma 4 31B, Gemini 2.5 Pro, Claude Sonnet 4.5 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
LMArena (text)	1441.5 ✎	▁▁▁▂▃▃▃▆▆▇█▃ beats 301/379	avg 1268	▇▇ 4.1%	blunt

math▁▂▃▅▆▅▄█▄▁▁▁ #55 of 398 · 16% of its score · 7 tests

Beats LongCat 2.0 (1.6T-A48B), Qwen3.6 Max Preview, Claude Sonnet 4.6 here, though they rank higher overall. Trails INTELLECT-3 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
IMO-AnswerBench lab +1	82.0% ✎	▁▁·▁▁▁▄▅█▄▂▃ beats 27/42	medium · avg 37%	▇▇ 4.3%	small run · several runs
HMMT February 2025 lab +1	97.1% ✎	▁▁▁▃▁▅▅▃██▃▄ beats 28/31	easy · avg 77%	▇ 3.5%	maxed out · small run · several runs
AIME 2025 lab	95.7% ✎	▁·▁▁▁▂▅▆█▇▅▃ beats 130/133	medium · avg 62%	▇ 2.8%	sharp · maxed out · small run
HMMT November 2025 lab +1	93.5% ✎	▁▁····▁▅▃█▇▁ beats 18/23	easy · avg 74%	▇ 2.4%	maxed out · small run · several runs
OTIS mock AIME 2024–2025	83.3% ✎	▃▃▄▃▃▃▇▇█▃▂▅ beats 120/193	very hard · avg 8.5%	▇ 2.4%	sharp · noisy
FormalProofBench (200 graduate-level Lean 4 problems, agentic)	4.5% ✎	▃▅·▃··▅▃█··▃ beats 2/10	very hard · avg 3.0%	▇ 0.7%	noisy
AIME 2026 I lab	92.9% ✎	█··········█ beats 1/1		—	not counted yet: no other model in the same run to compare it with

web search▂▁▂▆▄█▃▃▁▁·▁ #68 of 80 · 11% of its score · 4 tests

Beats DeepSeek V4 Flash (Preview, Apr 2026), Nemotron 3 Ultra 550B-A55B here, though they rank higher overall. Trails GPT-4.1 nano, GPT-4o mini (2024-07-18), Mistral Small 3.1 24B (2503) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
BrowseComp-ZH lab +1	66.6% ✎	▁▁▁▁▁▃·▁▄▅▄█ beats 18/23	hard · avg 22%	▇▇ 3.8%	at its level · small run · several runs
BrowseComp lab +1	52.0% ✎	▂·▄▁▂▃▆▄█▂▃▄ beats 20/42	hard · avg 18%	▇ 3.0%	at its level · small run · several runs
Humanity's Last Exam with tools lab +1	42.8% ✎	▂▂▂▁▄▂▃▆█▇▂▆ beats 19/54	hard · avg 16%	▇ 2.2%	at its level · small run · several runs
BrowseComp with context management lab +1	67.5% ✎	▅·▅▃·█▃·█▅·▃ beats 3/14	medium · avg 41%	▇ 2.1%	at its level · small run · several runs

other tests13 not in a task · 34% of its score

test	result	vs the others in its run	difficulty	counts	why
GPQA Diamond +2	83.3% ✎	▁▁▁▁▂▄▆▄▅▇█▅ beats 137/208	hard · avg 40%	▇▇▇▇▇ 12%	sharp · several runs
Humanity's Last Exam lab +1	24.8% ✎	▁▁▁▃▆▆▇██▆▁▃ beats 58/104	very hard · avg 4.9%	▇ 3.6%	small run · several runs
Medmarks-V weighted mean win rate (61 models, 71 configurations)	62.0% ✎	▂▃▅▂▆▃█▅█▅▃▆ beats 31/34	medium · avg 42%	▇ 2.9%	blunt · at its level
EQ-Bench 3 (emotional intelligence, judge Elo)	1271.3 ✎	▁·▂▁▄▅▅▃█▄▁▁ beats 39/62	avg 1044	▇ 2.5%	typical
Chess Puzzles (Epoch)	6.0% ✎	█▃▁▁▁▂▁▄▂▂▁▁ beats 47/135	very hard · avg 5.9%	▇ 2.0%	sharp · noisy
ResearchRubrics	62.0% ✎	███··██·█··█ beats 5/6	hard · avg 31%	▇ 1.9%	at its level
LongBench v2 lab	59.1% ✎	▂·▁·▂▂▁▃▄▁█▄ beats 19/28	very hard · avg 35%	▇ 1.9%	at its level · small run
RPC-Bench overall, text input (Z.ai) lab	28.8% ✎	▄··▄▄·▄▄█·▄▄ beats 3/8	very hard · avg 11%	▇ 1.8%	at its level

5 more tests

test	result	vs the others in its run	difficulty	counts	why
MMLU-Pro lab	84.3% ✎	▁▁·▁▁▁▃▂▄▅█▁ beats 163/188	medium · avg 56%	▇ 1.6%	small run
SimpleQA Verified	32.2% ✎	▁▁▃▂▃▃▆▆█▃▂▅ beats 22/78	very hard · avg 8.0%	▇ 1.5%	at its level
GAIA (no-file subset)	61.9% ✎	▃▃····█▃···▃ beats 0/6	medium · avg 47%	▇ 1.0%	at its level
CyberGym lab	23.5% ✎	▄···▄··▄█▄█▄ beats 0/8	hard · avg 19%	▇ 0.9%	at its level · small run
Vectara hallucination leaderboard (HHEM-2.3, factual consistency rate)	88.3% ✎	▁▂▁▂▅█▂▃▄▃▁▁ beats 29/98	very easy · avg 90%	▇ 0.0%	blunt

$ where to run it

Through an API

api · $1.00/M · $0.6 in / $2.2 out per 1M tokens · 205k context · docs.z.ai ↗ · price sourceStandard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ.

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 GLM-4.7 · needs ~214.7 GB · from huggingface.co/unsloth
chat in the terminal
llama cli -hf unsloth/GLM-4.7-GGUF:UD-Q4_K_XL
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf unsloth/GLM-4.7-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 GLM-4.7 · needs ~214.7 GB · from huggingface.co/unsloth
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/unsloth/GLM-4.7-GGUF@UD-Q4_K_XL -y
Then run lms ls to see its name, and lms chat followed by that name

MLX (Mac)

version

Install MLX (Mac) · Apple silicon Macs only, text only.
with uv
uv tool install mlx-lm
or
pip install mlx-lm
Run GLM-4.7 · needs ~198.6 GB, Mac only · from huggingface.co/mlx-community
chat in the terminal
mlx_lm.chat --model mlx-community/GLM-4.7-4bit
Or start an OpenAI-compatible server at http://localhost:8080/v1
mlx_lm.server --model mlx-community/GLM-4.7-4bit

Weights and documentation: huggingface.co/zai-org ↗

$ how the score was worked out · zai-org/GLM-4.7 model card 32%, GLM-5 Team (Zhipu AI) — GLM-5… 19%, zai-org/GLM-5 model card 15% 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

source	results	share of the score	licence
zai-org/GLM-4.7 model card · lab-reported	17	▇▇▇▇▇▇▇▇▇▇ 32%	MIT
GLM-5 Team (Zhipu AI) — GLM-5 technical report · lab-reported	10	▇▇▇▇▇▇ 19%	CC BY 4.0
zai-org/GLM-5 model card · lab-reported	8	▇▇▇▇▇ 15%	MIT
LMArena leaderboard dataset	2	▇▇▇ 11%	CC BY 4.0
Epoch AI — internal evaluations	4	▇▇▇ 11%	CC BY 4.0
Medmarks: A Comprehensive Open-Source LLM Benchmark Suite for Medical Tasks — authors' own runs of 61 models	1	▇ 3%	CC BY 4.0
stepfun-ai/Step-3.5-Flash model card	2	▇ 3%	Apache-2.0
EQ-Bench 3 canonical leaderboard results	1	▇ 3%	MIT
RPC-Bench leaderboard (Z.ai research paper comprehension) · lab-reported	1	▇ 2%	Apache-2.0
FormalProofBench (Vals AI, arXiv:2603.26996v1)	1	▇ 1%	CC-BY-4.0
Vectara hallucination leaderboard (vectara/hallucination-leaderboard repository)	1	▇ 0%	Apache-2.0
prior · pull toward the middle		▇ 0%
weighted average		= 68.7