whatmodel.sh / GLM-4 (0520)
GLM-4 (0520)
↗ share
+ contribute
GLM-4 (0520) is a closed-weights language model from Z.ai, released May 2024.
Catalogue identity notes
Comparison-only. Arena entry 'glm-4-0520' (text). LMArena's leaderboard table (arena-leaderboard space @6f2eeeb) labels 'glm-4-0520' as 'GLM-4-0520' (Zhipu AI, Proprietary), linking https://open.bigmodel.cn/dev/api#language. Released 2024-05: date in Zhipu's API id glm-4-0520 (May 2024 GLM-4 upgrade, per Zhipu's GLM tech report). Arena name equals the API id; no longer a public pay-as-you-go model on BigModel.
$ how it compares
The LLMs ranked either side of GLM-4 (0520), out of 828 with a score, then more from Z.ai.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 375 |  LFM2.5 8B-A1B | 52±1 | local | ||
| 376 |  GPT-4 (0314) | 52±1 | comparison only | ||
| 377 | GLM-4 (0520) | 52±1 | comparison only | ||
| 378 |  ERNIE 4.5 21B-A3B | 52±1 | comparison only | ||
| 379 |  Bielik 11B v2.2 Instruct | 52±3 | local | ||
| more from Z.ai | |||||
| 33 | GLM-5.3 | 76±1 | api $2.15/M +local | ||
| 52 | GLM-5.3-Flash | 74±1 | api $0.24/M +local | ||
| 57 | GLM-5.2 | 74±1 | api $2.15/M +local | ||
$ what it's good at
agent tasks▁▁▂▆█▆▅▄▄▃▂▁ #150 of 340 * · 2.0% of its score · 1 test
Beats DeepSeek V3.2 Exp, Qwen3 235B-A22B Thinking 2507, o4-mini here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Large here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| BFCL (version not stated) lab | 81.8% ✎ | █·········██ beats 4/5 | 2.0% |
coding█··········█ #1 of 2 * · 8.4% of its score · 1 test
chat & writing▁▁▁▃▅▄▆█▇▇▇▃ #297 of 463 · 56% of its score · 3 tests
Beats Mistral Small 4 (119B-A6.5B), Ministral 3 14B Reasoning 2512, Qwen2.5 Coder 32B Instruct here, though they rank higher overall. Trails Flan-T5 XXL, Granite 4.0 H 1B, Granite 4.0 1B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena-Hard (v0.1, GPT-4-1106 judge) | 63.8% ✎ | ▂▁▁▅█▅▄▃▃▅▃▁ beats 44/60 | 32% |
| Polyrating rating (bias-aware re-fit of Chatbot Arena votes, mid-2024) | 1202.0 ✎ | ▂▁▂▁▃▇█▆▇▄▂▂ beats 81/95 | 12% |
| LMArena (text) | 1273.4 ✎ | ▁▁▁▂▃▃▃▆▆▇█▃ beats 105/379 | 12% |
math█··········█ #1 of 2 * · 12% of its score · 2 tests
| test | result | vs the others in its run | counts |
|---|---|---|---|
| GSM8K lab | 93.3% ✎ | ▂▂▂▅▅▆▅▇█▇▄▃ beats 132/146 | 6.6% |
| MATH (subset unspecified) lab | 61.3% ✎ | ▂▄▄▃█▇▅▆▄▃▄▁ beats 35/80 | 5.1% |
other tests3 not in a task · 22% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| GPQA (subset unspecified) lab | 39.9% ✎ | ▂▁▂▂▄▃▆▃▅█▄▃ beats 37/104 | 8.2% |
| BIG-Bench Hard lab | 84.7% ✎ | ▁▁▁▁▂▄▅▅█▃▄▁ beats 117/129 | 7.8% |
| MMLU lab | 83.3% ✎ | ▁▁▁▁▁▂▃▅█▃▆▃ beats 148/183 | 5.6% |
$ where to run it
No verified API or download recorded yet · add one
$ how the score was worked out · ChatGLM: A Family of Large La… 44%, Arena-Hard-Auto v0.1 leaderbo… 32%, Polyrating: A Cost-Effective… 12% and 1 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 |
|---|---|---|
 ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools (Team GLM, arXiv 2406.12793v2) · lab-reported | 7 | 44% |
 Arena-Hard-Auto v0.1 leaderboard (repository README, pre-v2) | 1 | 32% |
| Polyrating: A Cost-Effective and Bias-Aware Rating System for LLM Evaluation (arXiv:2409.00696v3) | 1 | 12% |
 LMArena leaderboard dataset | 1 | 12% |
| prior · pull toward the middle | 1% | |
| weighted average | = 52.1 |
$ 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