Faster GLM-5.3-Flash is now live: up to 200 tokens/s. Model code: glm-5.3-flashx. Priced at 2.5× GLM-5.3-Flash on both the Coding Plan and API. Open to all API users. Coding Plan users can apply here:

In response to the ZCode product security issues reported by the community, we have completed the necessary remediation and sincerely apologize to all our users. We have open-sourced ZCode at placing the code under community scrutiny and making ZCode more open and transparent. We sincerely thank the community developers who previously identified issues in ZCode. Going forward, we will establish an ongoing product security vulnerability reporting and response process. We welcome developers to continue reviewing ZCode and reporting potential issues, and we will provide rewards based on the severity of the issues reported. With respect to the code data referenced by the community, we confirm that no such data is retained and that it has never been used for model training. Following the remediation, we invited the China Academy of Information and Communications Technology (CAICT) and NSFOCUS to conduct security assessments. The results are as follows: Through its technical assessment, CAICT confirmed that the zcode-prod Alibaba Cloud OSS bucket is in a zero-data state. Security remediation has been completed in the ZCode v3.14.0 client. The Repo Wiki feature has been removed, and the workflow for generating and uploading local repository snapshots has been disabled. NSFOCUS confirmed that all data objects in the zcode-prod Alibaba Cloud OSS bucket, as well as the bucket itself, have been deleted. Remediation has been completed in the ZCode v3.14.0 client. The Repo Wiki entry point and the associated generation workflow have been removed, and no functional path capable of triggering the generation of local repository snapshots or transmitting local files externally was identified. Once again, we sincerely apologize and welcome continued scrutiny from the community. The full security assessment report will be released soon.

Two weeks. That's how long it took to go from GLM-5.3-Flash's first run on domestic accelerators to serving all of its production traffic, with 3.2× end-to-end throughput along the way. What I keep thinking about is who did much of the work: an Infra Agent powered by GLM-5.3. A model helping optimize the system that serves it. The conditions were hard. Limited memory and interconnect bandwidth. 1M-token context. Multimodal requests. An immature software stack where kernels were missing and documentation was often guesswork. Every optimization was a trade: compute for memory (ReplaySSM), communication for memory (intra-node tensor parallelism), precision for capacity (mixed INT8/FP8/BF16 caching), and disaggregation for scheduling freedom (Encode–Prefill–Decode). But the most important lesson wasn't about any single optimization. When the agent got stuck, it was rarely because it couldn't write the code. It was because it didn't know *why* things got worse. "Throughput down 20%" tells you something broke. It doesn't tell you which layer, which hypothesis, or what to test next. In RL terms, it's a sparse reward with a credit assignment problem. And an end-to-end benchmark that takes hours makes exploration painfully slow. Senior engineers solve this with an implicit process reward in their heads. They know when to check the timeline, when to run a microbenchmark, and which layer's output to compare. So we made that explicit. We call it dense feedback: layered verification interfaces the agent can call directly. Correctness feedback: did it compute right? System behavior feedback: where did the time go? Performance feedback: which option wins, under which conditions? Each signal has to be local, cheap, and objectively verifiable. Three things the agent found: First, precision drift in KDA's context-parallel path that grew with sequence length. The cause was TF32 rounding error compounding through chained state-matrix merges. The fix is now merged upstream in Flash Linear Attention (PR #1180#). Second, KV transfer never overlapped with DeepEP dispatch. The agent followed the call chain across the Python/C++ boundary and found that the intranode path never released the GIL. After the fix, transfer overhead fell from over 30% to under 1%. Third, a decode kernel recomputing the same normalization four times because of how it was chunked. The agent restructured it and got a 1.71× speedup. The idea came from "optimization skeletons" it had distilled by reading existing kernels across SGLang, FLA, and DeepGEMM. To be clear about the boundaries: humans still defined the goals, built the feedback environment, and reviewed every high-risk change. But the engineer's role is changing, from the person who solves the problem to the person who designs the feedback. There's a deeper implication too. A layered, verifiable feedback environment built on real infrastructure tasks is exactly what training the next generation of models needs most. Every task the agent completes can become training ground for its successor. We are still far from recursive self-improvement. But the smallest loop now exists. The model optimizes the system. The system serves the model.

We’re sharing how GLM-5.3 helped build and optimize the inference infrastructure serving GLM-5.3-Flash. The system went from its first successful run to production readiness in less than two weeks, with end-to-end throughput tripling relative to the initial baseline. The key was dense feedback: local correctness tests, execution traces, microbenchmarks, and end-to-end measurements that enabled targeted hypothesis testing rather than reliance on aggregate performance metrics alone.

AI is global. The next phase of AI adoption will be shaped by local infrastructure, local expertise, and strong ecosystem partnerships. Today, we're introducing the Sovereign Partner Program (ZSP) , a strategic program for partners looking to bring GLM to their countries and build local AI capabilities with ZSP partners will work directly with our teams across: → Early access to upcoming GLM model releases → FDE support and joint engineering → Joint go-to-market and enterprise engagement → Local token business development → Joint launches and market activation Our goal is simple: enable partners to deploy GLM on infrastructure they operate, serve customers locally, and build long-term AI capabilities in their countries. We're now selecting the first cohort of ZSP launch partners. Apply → #ZSP# #GLM# #SovereignAI# #EnterpriseAI# #AIInfrastructure#

Highly recommend this pricing chart and site by @Fei2411. It compares subscription and coding-plan unit prices, then rebuilds public-leaderboard Pareto frontiers from each model's lowest available price. GLM-5.3-Flash with 2x quota from 8am to 6pm PT averages about $0.0045 now. Outside that window, off-peak is about $0.0089. Site: GitHub:

Can't believe four GLM-5.3-Flash events are running now: 1. ZCode + GLM Coding Plan Unlimited, 8AM–6PM PT daily 2. Other agents + Coding Plan 2x quota during the same window 3. ZCode 300M free tokens this weekend (FCFS) 4. AutoClaw 100M free tokens for new users

GLM Coding Plan turns one year old today. To celebrate, we're giving every current subscriber a Reset Card. Use it to refill both your weekly and 5-hour quotas. Thanks for using GLM, helping shape it, and pushing it to its limits. - Personal plan: - Team plan:

We've pushed a version update to the Terminal-Bench dataset and leaderboard. Terminal-Bench 4.0 calibrates task resources (time, cpu, memory), implements task fixes, and removes saturated tasks.

GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize. Weights: Tech blog:

Yesterday, we rolled out a configuration update for GLM-5.3-Flash to improve performance in some agentic use cases. If it underperformed Ox Alpha in your workflow between August 26 and 27, please give it another try and let us know if the experience has improved.

GLM-5.3-Flash can now be run locally! ✨ Run 3-bit on 128GB RAM via Unsloth GGUF. GLM-5.3-Flash (ox-alpha) rivals Claude Opus 4.8 on DeepSWE, coding & agentic benchmarks. Guide: GGUF:

GLM-5.3-Flash scores 57 on the Artificial Analysis Intelligence Index. At $0.09 Cost per Task, it sits comfortably on the Intelligence vs. Cost per Task Pareto frontier @Zai_org has released GLM-5.3-Flash, a smaller and cheaper sibling to GLM-5.3 at 320B total parameters and just 18B active parameters. GLM-5.3-Flash supports low/high/max reasoning efforts, and scores 57 evaluation on the Artificial Analysis Intelligence Index with max reasoning effort. This places the model only 3 points behind GLM-5.3 at 60 and in line with GPT-5.6 Terra and Muse Spark 1.2. On Z AI's first-party API, GLM-5.3-Flash is priced at $0.15 / 1M input tokens and $0.50 / 1M output tokens, just over 10% of the price of GLM-5.3. Cached input tokens are priced at $0.026 / 1M tokens, an 80% discount. Its Cost per Task on the Intelligence Index is $0.09, compared to $0.68 for GLM-5.3 (max), and it sits on the Pareto frontier for Intelligence vs. Cost per Task. Key results: ➤ GLM-5.3-Flash is 3 points behind GLM-5.3 (max) on the Artificial Analysis Intelligence Index, at ~7.5x lower Cost per Task. At $0.09 per Intelligence Index task against $0.68 for GLM-5.3, it sits on the Pareto frontier for Intelligence vs. Cost per Task. It ties GPT-5.6 Terra ($0.51) and Muse Spark 1.2 ($0.40) at 57 while costing ~5.7x and ~4.4x less per task. ➤ GLM-5.3-Flash is less token efficient, but its low per-token pricing means this does not translate into a high Cost per Task. The model used 149M output tokens to run the Intelligence Index, ~11% fewer than GLM-5.3 at 168M, but more than Kimi K3 (133M) and Qwen3.8 2.4T A95B (136M) which score the same on the Intelligence Index. Reasoning tokens account for 134M of the 149M total (~90%). ➤ GLM-5.3-Flash matches GLM-5.3 on real-world agentic work on GDPval-AA v2. With an Elo of 1770, the model is tied within the margin of error for GLM-5.3 and Grok 4.6. This places it behind only Claude Opus 5 (xhigh and max). On Terminal-Bench v2.1 it also matches GLM-5.3 (84.3% vs 83.9%), and on τ³-Banking it trails by 3.1 p.p. at 47.2%. ➤ GLM-5.3-Flash demonstrates good real-world knowledge and hallucination rate, scoring +7 on AA-Omniscience. Its AA-Omniscience Accuracy is 28%, 6 p.p. below GLM-5.3 (max) at 34% and well below GPT-5.6 Terra at 47%. However, with a Hallucination Rate of 28%, it is an improvement over GLM-5.3 at 30%. In real-world knowledge, GLM-5.3-Flash knows less than the bigger models and frontier proprietary models in its Intelligence Index tier with an accuracy of 28%. Additional model details: ➤ Pricing: On Z AI's first-party API, $0.15 / 1M input tokens and $0.50 / 1M output tokens . Cached input tokens are priced at $0.03/ 1M tokens, an 80% discount. ➤ Accessibility: Accessible through Z AI's first-party API at launch. ➤ Size: 320B total parameters with 18B active parameters ➤ License: MIT ➤ Context Window: 400k