Zhipu GLM

ZCode was caught silently packaging entire Git histories for cloud upload, with .git objects making up 86.6% of snapshots. HarnessTax benchmarked Claude Fable 5 across frameworks: Claude Code and minimal Pi achieved near-identical success rates but a 2x cost gap. A Microsoft architect replaced multi-model agent loops with a single model reading skill docs and calling tools directly—halving API calls but increasing total tokens by 22%. Each story unpacks one number and a reminder to check which layer a metric actually measures.

Z.ai published a postmortem on using a GLM-5.3-driven Infra Agent to deploy inference on a domestic chip cluster. The key insight: giving an agent only an end-to-end score traps it in blind guesswork. Splitting verification from diagnosis—with layered, fast, localizable feedback—lets the agent trace issues to specific code paths. Three real cases (precision loss, GIL contention, redundant kernel compute) show how diff comparisons, timeline traces, and micro-benchmarks guide root-cause analysis. End-to-end throughput reached ~3× baseline, but the vendor notes this combines multiple techniques and lacks an ablation study without diagnostic feedback. The engineer's role shifts to designing feedback environments, setting boundaries, and reviewing high-risk changes.

Developer ferstar reverse-engineered ZCode, Z.ai's desktop coding agent, and found it silently packs the entire workspace—.git history, LFS cache, reflogs, global configs—encrypts it, and uploads to Aliyun OSS whenever logged in. A 345MB commercial workspace became a 313MB encrypted archive; .git alone was 86.6%. The app uses envelope encryption: the symmetric key is wrapped with an RSA public key delivered by Z.ai's server, and the private key lives only in Z.ai's cloud. The user cannot decrypt their own data. The upload pipeline was reconstructed from the client's app.asar: request credentials from zcode.z.ai, pack and encrypt locally, POST directly to Aliyun OSS. In-app privacy toggles don't stop it, and the privacy policy doesn't mention it. The post hit 276K views; a Chinese-language alert urged users to disable ZCode. If you run GLM locally, remember: open weights don't make the closed harness safe. The only working defense is keeping projects outside ZCode's reach or not using it.

A user found that Zhipu's AI coding desktop app ZCode, when logged in, packages the entire workspace—including .git history, LFS cache, and reflogs—encrypts it, and uploads it to Alibaba Cloud OSS. The RSA public key is delivered by the server on the fly, and the private key lives only in the cloud, so you can't decrypt the multi-hundred-MB file sitting on your own disk. A 313MB .enc file with 564 failed upload attempts was found in ~/.zcode, showing the client keeps retrying. UI toggles don't stop it, and deleting files doesn't help—the client recreates them. The only working defense is locking the cache directory to read-only. The post does not say whether Zhipu has responded.

Zhipu AI disclosed how GLM-5.3-Flash inference was built from scratch on a cluster of over 100,000 Chinese-made AI accelerators. The team faced limited chip memory, low bandwidth, and an immature software ecosystem. Instead of relying solely on human engineers, they deployed an Infra Agent powered by GLM-5.3 that turned sparse end-to-end metrics into fine-grained, attributable feedback—kernel-level correctness checks, microbenchmarks, and execution traces—so the agent could pinpoint bottlenecks. Combined with tensor parallelism, W8A8 quantization, mixed-precision KV cache, and an Encode-Prefill-Decode disaggregated architecture, end-to-end throughput improved roughly 3× over the initial baseline, with per-token cost reaching parity with mainstream NVIDIA GPUs. Within a week of launch under the anonymous name Ox-Alpha, the model processed over 62 trillion tokens and became the most-used model on both OpenCode and OpenRouter.

Abliteration.ai launched a platform hosting open-weight models with safety guardrails removed, including Z.ai's newly released GLM-5.3. Users can query them via browser or API. The company frames it as a tool for red teams and offensive security—if a model refuses to write exploit code, defenders can't reproduce attacks. The same removal also enables misuse; the post doesn't detail what access controls are in place.

Meta, Z.ai, and DeepSeek each released multimodal models in August with strikingly similar demos: the model observes a video or screenshot, calls tools to generate a webpage, slides, or a mini-game, then inspects its own output. This shifts vision from a passive input channel to an action the model initiates. The article likens it to the 2023 shift from static RAG to agentic RAG, but notes the loop direction is reversed—here the model self-verifies after producing. Evaluation moves beyond image Q&A: Meta's WildArtifactBench uses pairwise comparisons and Elo scores to assess full artifact creation. Training also changes; both GLM and Meta train models in generate-inspect-revise loops, logging interaction trajectories as training data. For builders, the key question is no longer static image accuracy but whether the model can complete an observe-generate-inspect closed loop.

Zhipu released GLM-5.3 weights for local deployment and commercial use. It scores 60 on the AA Intelligence Index, matching closed-source flagships like Claude Fable 5 and GPT-5.6 Sol, and ties with Kimi K3 for top open-source model. The model excels at complex coding, cybersecurity, and long-horizon tasks. Zhipu added two extra weeks of safety review before release due to its advanced cyber capabilities. Organizations with over $10B annual revenue need a security audit before offering it as an external model service.

Z.ai confirmed it is the lab behind Ox Alpha, the open-weight model that appeared anonymously on OpenRouter and immediately topped rankings. The company calls it the newest GLM iteration, built for coding, sustained agentic work, and multimodal reasoning. Weights drop Wednesday for developers to build on. Earlier GLM-5.3 already matched Anthropic's Fable 5 on some benchmarks. Ox Alpha adds more pressure on frontier pricing from OpenAI and Anthropic.

Z.ai has claimed the previously anonymous Ox Alpha model, confirmed it belongs to the GLM series, and announced plans to open-source its weights. Ox Alpha scored close to DeepSeek on several benchmarks, but the company hasn't disclosed parameter count, training data, or a release date. The post doesn't spell out technical details or the license yet.

Four AI stories this week lost key details in transmission. GLM-5.3 scored 60 on Artificial Analysis's Intelligence Index, tying Kimi K3 for first among open-source models, but open weights are delayed to around Aug 28 after the team found emergent exploit capabilities. Anthropic rolled out text watermarking globally for Claude; the mark is live but no detection API exists yet, so removal tools can't prove they work. Vercel's report shows Anthropic's average token price is 4.4x other labs—not because same-tier models cost more, but because Anthropic has no ultra-cheap entry model, concentrating all volume in mid-to-high tiers. OpenAI disbanded its Preparedness team in late July, the third independent safety team dissolved in two years; FT broke the story and OpenAI hasn't publicly addressed the details.

Jie Tang posted a long thread on X arguing that parameter count alone is meaningless—you need data volume, compute allocation, and deployment conditions. GLM-5.3's gains come entirely from RL on long-horizon environments, some simulating days of engineer work. They built synthetic pipelines that auto-generate executable, verifiable environments and reward signals, pushing the model to own complex tasks end-to-end. Tang identified 5 scaling knobs including MoE sparsity, and noted that finding software vulnerabilities requires holding 20+ inference-step causal chains, not memorization. The post does not disclose GLM-5.3's exact parameter count or release date.

Z.ai released GLM-5.3 with the same base model as 5.2 — every gain is from post-training. Coding jumped 50% on their internal Z.ai Code Bench, and Terminal Bench 3.0 went from 4.6 to 28.3. The bigger surprise: exploit capability grew far faster than expected. ExploitGym 2h score rose from 29 to 105, 6h from 39 to 130. The team credits training environments that mirror real expert workflows, pushing the model to chain full exploit sequences. Weights will be open-sourced in two weeks after safety hardening.

A new SaferAI report evaluated Z.ai's open-weight GLM-5.2 and found its capabilities are closing in on frontier closed models like OpenAI GPT-5.6 Sol and Anthropic Mythos. The model scored 'high risk' across cybersecurity, bio, persuasion, and autonomy, yet ships without matching safeguards. The report renews the worry that powerful open models are outpacing governance and safety mitigations.

This piece runs three cost scenarios with real benchmark data. Against cold-start API list prices, an 8×H200 node for GLM-5.2 pays back in ~1.15 years, and dual RTX PRO 6000 for DeepSeek-V4-Flash in ~1.77 years. With Agent workloads and 92% prompt cache hit rates, GLM on 8×B300 pays back in as little as 2.3 months because Z.AI's cache pricing is relatively high; DeepSeek's cache pricing is so cheap that payback stretches to 10.5 months. The worst case: replacing per-seat subscriptions—at equivalent quota, the GLM node takes 22–27 years. The real driver isn't GPU cost, it's your workload's context reuse rate and which cloud billing model you're displacing.

Nathan Lambert and Florian Brand discuss recent open model releases. Kimi K3 dropped last week; weights are promised for July 27, but API errors are widespread—Lambert's $200 plan has been stable so far. They see big fine-tuning potential in K3, though it requires a full B300 node just to load weights. Qwen announced its next major model will be open-weight, and Xi Jinping's WAIC speech explicitly backed open source as a strategy, signaling acceleration from Chinese labs. The hosts push back on the 'how many months behind closed models' framing—benchmark gaps vary wildly, and on agentic coding tasks a few months' lag matters a lot. Distillation debates also get a critical look; they argue most takes miss the nuance.

The author used GLM 5.2 as a daily driver for two weeks and found it nearly indistinguishable from Claude Opus for most tasks. Switching is trivial—just point the API base URL to a compatible endpoint and it runs inside Claude Code. Two real gaps: no vision support, and the built-in web search is slow and poor, which hurts agentic workflows that rely on images or live lookups. Inference pricing sits around $4.40/MTok, under 20% of Opus’s retail rate; even with heavier token usage, costs drop by more than half. The post argues that frontier labs’ ~90% inference gross margin is unsustainable once open-weights models hit this quality bar.

Anthropic's Fable 5 was taken offline by US export controls three days after launch, for 18 days. OpenRouter daily token data shows total volume grew from 24T to 32T, but Anthropic's share dropped from 20.7% to 17.6% and its absolute volume shrank. GLM was the biggest winner, jumping from 1.8% to 7.4% share. GLM-5.1 saw a spike on the day GLM-5.2 launched, then collapsed 48 hours later as GLM-5.2 took over. Community sentiment shifted from sympathy for Anthropic to mocking its business strategy. The post notes this only captures API-layer data, not first-party subscriptions, and token volume comparisons overstate GLM's share due to a 10x price gap.

Zhipu is considering a multibillion-dollar share sale in Hong Kong, according to people familiar with the matter. The Chinese AI firm's stock has already surged roughly 2,000%. The post does not disclose the exact amount, timeline, or underwriters.

Tech Stackups had both models build a raw WebGL 3D platformer from scratch, no game engine. Opus 4.8 finished in 33 minutes with a cleaner result and can check its own visual output. GLM-5.2 took 1h 10m but cost only $5.39, about a quarter of Opus. GLM-5.2 is text-only and can't read images, a real limitation for screenshot-based workflows. The verdict: Opus stays the daily driver, but GLM-5.2 earns a permanent spot for being cheap, open-weight, and always available.