AITOPAI热报

GLM-5.3

Zhipu AI·China·Released 2026-02-12
Open weightsGeneral LLMReasoningTool use

Zhipu AI's flagship LLM for long-horizon coding, agents, and complex project delivery with a 1M token context window.

1Mcontext
$1.2 / $4.18per 1M tokens (in / out)
753Bparameters
162stories in our feed

Specs

hover a value to see its source
Context window1M
Max output128K
Parameters753B
Knowledge cutoff—
Licenceother
Commercial useRestricted
InputText
OutputText
HF downloads (30d)2.1M
Familyglm

API price

USD per 1M tokens
Input$1.2
Output$4.18
Cached input$0.299
Change history
2026-10-21Price $1.4/$4.4 → $1.2/$4.18models.dev

News timeline

from the AITOP feed, one story per eventAll 120 →
2026-11-06
Revolut 自建 GPU 跑 GLM 替代 Claude Code，成本降 90%Blognone
2026-11-02
Z .ai 开源模型 GLM 5.3 可独立挖掘 Firefox 0day 并构建攻击链rohanpaul_ai
2026-10-30
开源模型 GLM 5.3 在 48 小时内自主复现 Firefox 漏洞利用链Decoder
2026-10-18
SemiAnalysis 播客拆解 GLM 5.3：后训练提升与国产芯片全栈适配shao__meng
2026-10-16
GLM-5.3 tops Epoch AI Cyber ECI among open-weight modelsEpoch AI
Anthropic 推出三层反蒸馏防御，Claude 推理 token 窃取量下降 90%–99%shao__meng
Meta 发布开源权重模型 Muse Spark Open：816B 总参数，登顶美国开源模型Artificial Analysis
2026-10-14
llm-gemini 0.35 支持 Gemini 4 Argon 模型Simon Willison’s Weblog
2026-10-06
Anthropic开放Claude Mythos 5.1网络安全访问rohanpaul_ai
Mistral Large 4 在 DeepSWE 基准测试中领先 GLM-5.3TestingCatalog
2026-10-02
GLM-5.3 matches Claude in vulnerability testingDeepLearning.AI
2026-09-30
OpenAI支持企业直接使用GLM和Kimi模型Xianbao Qian
智谱GLM-5.3编写网络攻击能力接近Claude MythosDecoder
OpenAI企业套餐加入GLM 5.3后发布开源模型安全报告orange.ai
Anthropic实测：GLM-5.3写攻击程序能力接近MythosLxfater
Anthropic测评GLM-5.3：开源模型达前沿水平shao__meng
Anthropic测试GLM-5.3：可写攻击程序，防护易绕过宝玉
2026-09-29
Anthropic 评测 GLM-5.3 安全性能kimmonismus
GLM-5.3在控制流劫持测试中取得突破Teortaxes
GLM 5.3 Flash 在 ARC-AGI 基准测试中表现优异ARC Prize
One-Shot OPD Method Reduces Training DataOpenBMB
开放模型在评测中访问外部世界AI21 Labs
2026-09-24
Artificial Analysis 推出 Terminal-Bench-Science 0.1 智能体科研基准榜单Artificial Analysis
GLM-5.3 Edges Space Bunny Alpha in Physics Simulation Testrohanpaul_ai
2026-09-17
Bolt 推出支持 GLM 5.3 Flash 等开源模型，使用量提升 50 倍bolt.new
Zai.org optimizes GLM-5.3-Flash for productionshao__meng
Zhipu optimizes GLM-5.3 inference with its own AI system宝玉
2026-09-07
阿里多模型入榜前端开发榜，wan3.0冲进视频榜前三IT之家
2026-09-06
Abliteration.ai提供移除安全防护的AI模型服务Decoder
2026-09-01
Z.ai GLM-5.3在CyberGym基准测试达84.5%DeepLearning.AI
Anonymous AI Model Audit ProtocolarXiv cs.AI
Zhipu Reports 2026 Mid-Year Resultsshao__meng
2026-08-31
GLM-5.3 通过微调达到开放模型最高分DeepLearning.AI
2026-08-29
GLM 5.3 models on Ollama cloudollama
GLM 5.3 在 Fireworks 上线Fireworks AI
2026-08-28
GLM 5.3 现已在 Perplexity Computer 上线Perplexity
Ollama上线GLM-5.3模型ollama
Ox Alpha身份揭晓：GLM-5.3 FlashAI Will
2026-08-27
智谱GLM-5.3 Flash跑10万张国产芯片Geek
2026-08-26
智谱开源 GLM-5.3-FlashIT之家