Gemini 4 Argon
Claude Opus 5.5
Muse Wonder
MiMo-V2.6-Pro
Qwen3.8-Max
Kimi K3
GPT-5.5
ERNIE 5.1
DeepSeek-V4.1-Flash-20260910
Grok 4.20 Beta
amazon-nova-experimental-chat
Dola Seed 2.0 Pro
NVIDIA-Nemotron-3-Ultra-550B-A55B
Gemma 4 31B
Step 5 Preview
GLM-5.3
GLM-5.3 is an open-weight model from Z.ai, released Aug 18, 2026. It ranks #27 of 351 on the LMArena text leaderboard with a rating of 1474, and #10 of 27 on the Vals Index with 59.73%. It costs $0.079 per million input tokens and $1.40 per million output tokens, with a 1M-token context window.
- Developer
- Z.ai (Zhipu AI)
- Released
- Aug 18, 2026
- Context window
- 1M
- Price in / out, per 1M tokens
- $0.079 / $1.40
- Weights
- Open · MIT
- Inputs
- Text
- Reasoning
- Yes
- Tool use
- Yes
- Hugging Face downloads, 30 days
- 2M
- Hugging Face likes
- 2.3K
Benchmarks
| Benchmark | Score bar | Accuracy | Rank | Cost / test |
|---|---|---|---|---|
| Vals IndexVals AI | 59.73% | 10 / 27 | $8.52 | |
| Terminal-Bench 4.0Coding | 38.89% | 13 / 55 | $9.37 | |
| SWE-benchCoding | 95.40% | 6 / 85 | $0.34 | |
| Vibe Code BenchCoding | 78.13% | 31 / 115 | $12.45 | |
| Code MigrationCoding | 55.07% | 11 / 28 | $34.91 | |
| LiveCodeBenchCoding | 80.53% | 68 / 143 | $1.4 / $4.4 | |
| Finance AgentFinance | 65.00% | 9 / 31 | $1.03 | |
| Finance IndexFinance | 60.54% | 8 / 31 | $2.21 | |
| Tax Agent BenchFinance | 39.70% | 10 / 76 | $1.80 | |
| Legal IndexLaw | 74.58% | 4 / 28 | $2.68 | |
| Legal Research BenchLaw | 49.04% | 9 / 85 | $2.24 | |
| MedCodeMedicine | 42.86% | 49 / 107 | $1.4 / $4.4 | |
| GPQA DiamondReasoning | 88.13% | 36 / 138 | $1.4 / $4.4 | |
| MMLU-ProReasoning | 86.77% | 38 / 138 | $1.4 / $4.4 | |
| Terminal-Bench ScienceReasoning | 5.71% | 22 / 42 | $15.63 |
Vals AI, as of Oct 24, 2026.
LMArena, by category
| Leaderboard | Rating | Rank | Votes |
|---|---|---|---|
| TextLMArena | 1474 | 27 / 351 | 25.1K |
| CodingLMArena | 1498 | 27 / 346 | 6K |
| MathLMArena | 1498 | 19 / 339 | 1.1K |
| Creative writingLMArena | 1459 | 25 / 349 | 5.7K |
| Instruction followingLMArena | 1480 | 19 / 351 | 8.4K |
| Hard promptsLMArena | 1490 | 21 / 351 | 15.8K |
| Multi-turnLMArena | 1470 | 37 / 349 | 3.7K |
| Long queriesLMArena | 1483 | 21 / 330 | 11.4K |
| ExpertLMArena | 1514 | 19 / 305 | 2.4K |
| Legal and governmentIndustries | 1473 | 39 / 325 | 2.2K |
| Medicine and healthcareIndustries | 1482 | 25 / 325 | 1.9K |
| Business and financeIndustries | 1459 | 34 / 344 | 4.6K |
| ChineseLanguages | 1532 | 16 / 328 | 1.9K |
| SpanishLanguages | 1461 | 41 / 249 | 818 |
| GermanLanguages | 1486 | 19 / 261 | 402 |
| FrenchLanguages | 1499 | 14 / 252 | 862 |
| JapaneseLanguages | 1458 | 28 / 236 | 457 |
| WebDevLMArena | 1623 | 25 / 124 | 10.3K |
LMArena, CC BY 4.0, as of Oct 22, 2026. Where a model is rated at several effort levels, its best is shown.
Compare GLM-5.3
Z.ai (Zhipu AI)
A Tsinghua spin-out that builds the GLM family of open bilingual models. Founded 2019, based in Beijing. z.ai
Questions
Who makes GLM-5.3?
GLM-5.3 is made by Z.ai (Zhipu AI), founded 2019 and based in Beijing.
How much does GLM-5.3 cost?
$0.079 per million input tokens and $1.40 per million output tokens, at its provider’s list price.
What is GLM-5.3’s context window?
1M tokens (1,048,576).
How does GLM-5.3 rank?
#27 of 351 on LMArena’s text leaderboard, rated 1474 (95% interval 1469–1479) from 25,144 votes, as of Oct 22, 2026.
Is GLM-5.3 open source?
GLM-5.3 has open weights under the MIT license, downloadable from Hugging Face (zai-org/GLM-5.3).