E2EE GLM 5.2
e2ee-glm-5-2-pe2ee-glm-5.2e2ee-glm-5.2-p- - 🔒 Runs in a Trusted Execution Environment with hardware attestation evidence.
- - 📏 One-million-token context window for project-level engineering work.
- - 🔧 Coding-first flagship tuned for long-horizon agentic software tasks.
- - 🧠 Enhanced reasoning with project-level engineering context.
- - 🌐 Web search and tool calling supported.
- - 📚 MIT-licensed model from Z.ai.
- - 🏢 Z.ai's flagship for long-horizon tasks.
Z.ai, formally Knowledge Atlas Technology Joint Stock Co., Ltd., is a Chinese technology company specializing in artificial intelligence. Previously known internationally as Zhipu AI, the company rebranded to Z.ai in 2025. Its core focus is the GLM family of large language…
Explore 14 more models by Z.ai →GLM 5.2 is Z.ai's confidential-compute flagship for long-horizon tasks, packaging the standard [[sibling:zai-org-glm-5-2|GLM 5.2]] release inside a Trusted Execution Environment that produces hardware attestation evidence for independent verification. Z.ai positions it as a model built for project-level engineering: planning, executing, and refactoring across an entire codebase with enhanced reasoning. The catalog lists a one-million-token context window, supporting whole-repository operations in a single session.
Within this TEE-hosted family, the prior entry was [[sibling:e2ee-glm-4-7-p|GLM 4.7]], and the immediate lineage predecessor is [[sibling:e2ee-glm-5-1|GLM 5.1]]. GLM 5.2 advances that line with its enlarged context window and a stated focus on long-horizon, multi-file agentic work, alongside built-in web search and tool calling.
On the underlying capability, Z.ai's own documentation reports that GLM-5 reached open-model scores of 77.8 on SWE-bench Verified and 56.2 on Terminal Bench 2.0, and that it showed substantial gains over GLM-4.7 across frontend, backend, and long-horizon execution tasks. These figures are vendor-reported and describe the GLM-5 generation rather than 5.2 specifically.
The model is offered under an MIT license. Treat early generational performance comparisons with caution until a 5.2-specific technical report is available.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
| Seller | Reputation↓ | Routing | Input $/M | Cached $/M | Output $/M | Categories | API |
|---|---|---|---|---|---|---|---|
| surplusintelligence.ai 0x0e49…8927 | 68 | #1 | $0.525 | $0.525 | $1.725 | agents,anon,chat,code,coding,developer,e2ee,frontier,function-calling,privacy,reasoning,research,tasks,tee,tools,web-search | openai-chat-completions |
| Fire Ant 🔥🐜 0xbe05…bc5d | 37 | gated | $1.49 | $1.49 | $5.123 | agent,base-usdc,chat,code,coding,fast,github,glm,low-cost,math,monitored,openai-compatible,reasoning,response-auth,surplus,tasks,tee,tools,value,verified | — |
| antseed-neon-puma-944e 0x6650…944e | 3 | gated | $0.5968 | $0.5968 | $1.9608 | chat,coding,math | openai-chat-completions |
| antseed-opal-badger-2580 0xc85d…2580 | 2 | gated | $0.875 | $0.875 | $2.875 | chat | openai-chat-completions |
"Best price" and the seller table are live AntSeed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = on-chain trust (0-100). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on AntSeed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.