{"id":45215777,"url":"https://github.com/zai-org/GLM-5","last_synced_at":"2026-03-05T15:01:15.171Z","repository":{"id":337971280,"uuid":"1153372704","full_name":"zai-org/GLM-5","owner":"zai-org","description":"GLM-5: From Vibe Coding to Agentic Engineering","archived":false,"fork":false,"pushed_at":"2026-02-12T07:13:55.000Z","size":3359,"stargazers_count":423,"open_issues_count":6,"forks_count":32,"subscribers_count":2,"default_branch":"main","last_synced_at":"2026-02-12T16:45:41.627Z","etag":null,"topics":["agentic-ai","coding","glm","llm"],"latest_commit_sha":null,"homepage":"https://z.ai/blog/glm-5","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/zai-org.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-02-09T08:17:02.000Z","updated_at":"2026-02-12T16:45:33.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/zai-org/GLM-5","commit_stats":null,"previous_names":["zai-org/glm-5"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/zai-org/GLM-5","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zai-org%2FGLM-5","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zai-org%2FGLM-5/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zai-org%2FGLM-5/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zai-org%2FGLM-5/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/zai-org","download_url":"https://codeload.github.com/zai-org/GLM-5/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zai-org%2FGLM-5/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30132585,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-05T14:41:47.141Z","status":"ssl_error","status_checked_at":"2026-03-05T14:41:21.567Z","response_time":93,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agentic-ai","coding","glm","llm"],"created_at":"2026-02-20T17:00:40.787Z","updated_at":"2026-03-05T15:01:15.144Z","avatar_url":"https://github.com/zai-org.png","language":null,"funding_links":[],"categories":["Others","Text","🏭 Industrial / Production Model Reports","2. Open Foundation Models"],"sub_categories":["Models","🔁 Iterative Self-Bootstrapping"],"readme":"# GLM-5\n\n\u003cdiv align=\"center\"\u003e\n\u003cimg src=resources/logo.svg width=\"15%\"/\u003e\n\u003c/div\u003e\n\u003cp align=\"center\"\u003e\n    👋 Join our\u003ca href=\"resources/WECHAT.md\" target=\"_blank\"\u003e Wechat\u003c/a\u003e or \u003ca href=\"https://discord.gg/QR7SARHRxK\" target=\"_blank\"\u003eDiscord\u003c/a\u003e community.\n    \u003cbr\u003e\n    📖 Check out the GLM-5 \u003ca href=\"https://z.ai/blog/glm-5\" target=\"_blank\"\u003etechnical blog\u003c/a\u003e.\n    \u003cbr\u003e\n    📍 Use GLM-5 API services on \u003ca href=\"https://docs.z.ai/guides/llm/glm-5\"\u003eZ.ai API Platform. \u003c/a\u003e\n    \u003cbr\u003e\n    👉 One click to \u003ca href=\"https://chat.z.ai\"\u003eGLM-5\u003c/a\u003e.\n\u003c/p\u003e\n\n## Introduction\n\nWe are launching GLM-5, targeting complex systems engineering and long-horizon agentic tasks. Scaling is still one of the most important ways to improve the intelligence efficiency of Artificial General Intelligence (AGI). Compared to GLM-4.5, GLM-5 scales from 355B parameters (32B active) to 744B parameters (40B active), and increases pre-training data from 23T to 28.5T tokens. GLM-5 also integrates DeepSeek Sparse Attention (DSA), largely reducing deployment cost while preserving long-context capacity.\n\nReinforcement learning aims to bridge the gap between competence and excellence in pre-trained models. However, deploying it at scale for LLMs is a challenge due to the RL training inefficiency. To this end, we developed [slime](https://github.com/THUDM/slime), a novel **asynchronous RL infrastructure** that substantially improves training throughput and efficiency, enabling more fine-grained post-training iterations. With advances in both pre-training and post-training, GLM-5 delivers significant improvement compared to GLM-4.7 across a wide range of academic benchmarks and achieves best-in-class performance among all open-source models in the world on reasoning, coding, and agentic tasks,  closing the gap with frontier models.\n\n![bench](resources/bench.png)\n\nGLM-5 is purpose-built for complex systems engineering and long-horizon agentic tasks. On our internal evaluation suite CC-Bench-V2, GLM-5 significantly outperforms GLM-4.7 across frontend, backend, and long-horizon tasks, narrowing the gap to Claude Opus 4.5.\n\n![realworld_bench](resources/realworld_bench.png)\n\nOn [Vending Bench 2](https://andonlabs.com/evals/vending-bench-2), a benchmark that measures long-term operational capability, GLM-5 ranks \\#1 among open-source models. Vending Bench 2 requires the model to run a simulated vending machine business over a one-year horizon; GLM-5 finishes with a final account balance of $4,432, approaching Claude Opus 4.5 and demonstrating strong long-term planning and resource management.\n\n![vending_bench](resources/vending_bench.png)\n\n## Download Model\n\n| Model     | Download Links                                                                                                                  | Model Size | Precision |\n|-----------|---------------------------------------------------------------------------------------------------------------------------------|------------|-----------|\n| GLM-5     | [🤗 Hugging Face](https://huggingface.co/zai-org/GLM-5)\u003cbr\u003e [🤖 ModelScope](https://modelscope.cn/models/ZhipuAI/GLM-5)         | 744B-A40B  | BF16      |\n| GLM-5-FP8 | [🤗 Hugging Face](https://huggingface.co/zai-org/GLM-5-FP8)\u003cbr\u003e [🤖 ModelScope](https://modelscope.cn/models/ZhipuAI/GLM-5-FP8) | 744B-A40B  | FP8       |\n\n## Serve GLM-5 Locally\n\n### Prepare environment\n\nvLLM, SGLang, and xLLM all support local deployment of GLM-5. A simple deployment guide is provided here.\n\n+ vLLM\n\n    Using Docker as:\n\n    ```shell\n    docker pull vllm/vllm-openai:nightly \n    ```\n\n    or using pip:\n\n    ```shell\n    pip install -U vllm --pre --index-url https://pypi.org/simple --extra-index-url https://wheels.vllm.ai/nightly\n    ```\n\n    then upgrade transformers:\n\n    ```\n    pip install git+https://github.com/huggingface/transformers.git\n    ```\n\n+ SGLang\n\n    Using Docker as:\n  \n  ```bash\n    docker pull lmsysorg/sglang:glm5-hopper # For Hopper GPU\n    docker pull lmsysorg/sglang:glm5-blackwell # For Blackwell GPU\n    ```\n\n### Deploy\n\n+ vLLM\n\n    ```shell\n    vllm serve zai-org/GLM-5-FP8 \\\n         --tensor-parallel-size 8 \\\n         --gpu-memory-utilization 0.85 \\\n         --speculative-config.method mtp \\\n         --speculative-config.num_speculative_tokens 1 \\\n         --tool-call-parser glm47 \\\n         --reasoning-parser glm45 \\\n         --enable-auto-tool-choice \\\n         --served-model-name glm-5-fp8\n    ```\n\n    Check the [recipes](https://github.com/vllm-project/recipes/blob/main/GLM/GLM5.md) for more details.\n\n+ SGLang\n\n    ```shell\n    python3 -m sglang.launch_server \\\n      --model-path zai-org/GLM-5-FP8 \\\n      --tp-size 8 \\\n      --tool-call-parser glm47  \\\n      --reasoning-parser glm45 \\\n      --speculative-algorithm EAGLE \\\n      --speculative-num-steps 3 \\\n      --speculative-eagle-topk 1 \\\n      --speculative-num-draft-tokens 4 \\\n      --mem-fraction-static 0.85 \\\n      --served-model-name glm-5-fp8\n    ```\n  \n    Check the [sglang cookbook](https://cookbook.sglang.io/autoregressive/GLM/GLM-5) for more details.\n\n+ xLLM and other Ascend NPU\n\n    Please check the deployment guide [here](https://github.com/zai-org/GLM-5/blob/main/example/ascend.md).\n\n## Citation\n\nOur technical report is coming soon.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fzai-org%2FGLM-5","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fzai-org%2FGLM-5","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fzai-org%2FGLM-5/lists"}