Zhipu AI released GLM-5.3 on August 14, 2026, an updated version of its open-weight coding model that uses the same base architecture as GLM-5.2. Every reported gain comes from extended post-training rather than a new pretraining run. The company describes GLM-5.3 as the strongest open-weight model for coding and highlights an emergent cybersecurity capability that grew as post-training scaled.<\/p>\n
GLM-5.3 shares the base model with GLM-5.2. Zhipu kept scaling the post-training stack it built for GLM-5.2, which includes IndexShare for efficient long-context processing, SAO for reinforcement learning on long-horizon tasks, and the open-source slime framework for large-scale asynchronous training. The improvements for GLM-5.3 come from adding more environments, more diverse tasks, and more compute spent training on them.<\/p>\n
The company reports a 50% improvement over GLM-5.2 on its in-house Z.ai Code Bench and open-source state of the art on public benchmarks including Terminal Bench 3.0 and Agents’ Last Exam.<\/p>\n
On Terminal Bench 3.0, GLM-5.3 jumps from 4.6 to 28.3. On DeepSWE v1.1, it rises from 46.2 to 66.9. On Agents’ Last Exam ALE-CLI, it improves from 23.8 to 28.5. Against a set of competing models, the posted numbers are:<\/p>\n
Zhipu pushed environment scaling toward tasks that look less like coding exercises and more like real units of expert work. The training environments cover a broader range of production workflows, with some tasks designed around how engineering and research work is actually carried out. Some represent several days of work for an experienced engineer. In one ML infrastructure task the model receives the same working environment as an engineer, with access to compute clusters, storage systems, internal documentation, codebases and experiment results, and must diagnose bottlenecks across the training stack, implement optimizations, run experiments and deliver a measurable end-to-end speedup while preserving correctness.<\/p>\n
Zhipu built pipelines that synthesize environments end to end, and for a subset of tasks the reinforcement learning reward signal as well. Research agents collect task patterns from real work and turn them into runnable long-horizon environments with multi-step dependencies and hidden state. A judge agent then attempts each task to verify it is actually solvable. Verifiers are synthesized without access to the reference solution, while solver trajectories are used to discover and close reward shortcuts.<\/p>\n
At Max effort, GLM-5.3 reaches 34.5% at roughly 75K output tokens per task, compared with 23.4% at 96K for GLM-5.2. At High effort, GLM-5.3 reaches 31.4% at around 50K output tokens, surpassing Claude Opus 4.8 at 29.5% with 120K. GLM-5.3 remains behind Claude Fable 5, which reaches 39.5% at Max effort.<\/p>\n
Zhipu introduced vulnerability discovery data and environments into the training mix, and the capability continued to develop as training scaled. According to the company, GLM-5.3 began to reason across multiple stages of exploitation, forming coherent plans for complete exploitation chains.<\/p>\n
On CyberGym, which starts from white-box source code and tests whether the model can identify and validate vulnerabilities by triggering faults, GLM-5.3 scores 84.5%, up from GLM-5.2’s 77.2%, the best result on the benchmark, ahead of Mythos 5 at 83.8% and GPT-5.6 Sol at 83.6%. On ExploitBench, which requires deeper reasoning about real vulnerabilities and their exploitation, GLM-5.3 reaches 54.4%, more than doubling GLM-5.2’s 24.4%, while Mythos 5 and GPT-5.6 Sol score 78.0% and 76.5%. On ExploitGym, which measures how many exploitation tasks a model can complete under time-normalized budgets, GLM-5.3 completes 105 tasks within two hours and 130 within six hours, compared with 29 and 39 for GLM-5.2. Mythos 5 completes 181 and 247 tasks. The pattern is consistent: the further up the exploitation chain a benchmark sits, the larger the gain over GLM-5.2 and the wider the remaining gap to the closed frontier.<\/p>\n
Since GLM-5.2, Zhipu has worked with several security teams in China to run its models against real-world codebases. After expert review, screening and deduplication, the model identified 2,436 vulnerabilities across 269 open-source projects, including 1,097 medium-to-high severity issues. The findings span system kernels, operating systems, browser engines, open-source infrastructure, web applications and network protocols. Many had remained unnoticed for years or decades, with the oldest dating back roughly 40 years.<\/p>\n
A public registry at cvd.z.ai tracks the findings. As of the published counters, it lists 2,436 findings tracked, 53 publicly disclosed, 2,383 under embargo, 1,097 critical and high, 269 open-source projects, and 45 years of impact. Severity distribution: 107 Critical, 990 High, 1,286 Medium, 53 Low. The oldest flaw was introduced in 1981, and on average a vulnerability lived 26.6 years before discovery. For disclosed issues the registry records the affected project, severity, CVE where available, and how long the vulnerability had remained in the codebase.<\/p>\n
All of this runs on slime, Zhipu’s open-source post-training framework for reinforcement learning scaling, with Megatron on the training side and SGLang on the rollout side. Its design keeps training, rollout and the data buffer on a single dataflow, so math, code, sandboxes, verifiers and long-horizon agentic environments plug in as data generation rather than as changes to the training loop. Additions through GLM-5.3 include top-p mask, top-k and full-vocabulary OPD, and configurations improving training-rollout consistency including R3-style setups and full numerical alignment between training and rollout paths. In the training-rollout consistency evaluation the average difference in log probabilities was controlled at the 1e-7 level, a reduction of more than 99.99% compared with previous setups.<\/p>\n
GLM-5.3 is available now through the GLM Coding Plan at z.ai\/subscribe and works with coding agents including ZCode, Claude Code and OpenCode. Weights are set to go open source two weeks after launch, once safety evaluation and hardening are complete.<\/p>\n
GLM-5.3 is an open-weight coding model released by Zhipu AI on August 14, 2026. It shares the same base model as GLM-5.2, with all reported gains coming from extended post-training.<\/p>\n
Zhipu reports that GLM-5.3 identified 2,436 vulnerabilities across 269 open-source projects, including 1,097 medium-to-high severity issues. The oldest flaw dates back about 40 years. The findings are tracked in a public registry at cvd.z.ai.<\/p>\n
Zhipu plans to release the GLM-5.3 weights two weeks after launch, once safety evaluation and hardening are complete.<\/p>\n