GLM 4.7
glm-4-7glm-4.7zai-org-glm-4-7zai-org-glm-4.7- 🆕 Z.ai's December 2025 update to the open-source GLM line.
- 🧠 Mixture-of-experts design with 358 billion total parameters.
- 📏 Roughly 200K-token context for long codebases and documents.
- 🔧 Built for agentic coding, tool use, and terminal tasks.
- 🎯 Vendor reports 42.8% on Humanity's Last Exam, up 12.4 points over GLM-4.6.
- 💬 Turn-level thinking control toggles reasoning depth per request.
- 🔒 Released under the permissive MIT license with open weights.
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 17 more models by Z.ai →GLM 4.7 is a large language model from Z.ai (formerly Zhipu AI), released in December 2025 as part of the company's open-weight GLM family. It is a multilingual Mixture-of-Experts model reported at 358 billion total parameters, optimized for coding, multi-step reasoning, and tool use. The weights ship under the MIT license, and the model supports an extended context window of roughly 200K tokens.
Building directly on GLM 4.6, Z.ai positions GLM 4.7 as a more engineering-focused step forward, strengthening agentic coding, terminal-based tasks, UI generation, and stable long-horizon execution. On the provider's reported Humanity's Last Exam benchmark it scores 42.8%, a 12.4-point gain over GLM 4.6, and the model card cites improved results on the τ²-Bench tool-invocation and BrowseComp web-browsing evaluations. A central new capability is finer control over reasoning: GLM 4.7 extends the family's interleaved thinking, letting developers disable reasoning for lightweight requests and enable it for complex tasks.
The release is accompanied by a lighter variant, GLM 4.7 Flash, aimed at efficiency-balanced deployment. GLM 4.7 sits between GLM 4.6 and Z.ai's subsequent generation, including GLM 5 and the newest GLM 5.2. The open weights are distributed through Hugging Face for local serving, and the model is also available via Z.ai's API.
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 |
|---|---|---|---|---|---|---|---|
| Venice.ai Proxy 0x1f22…18c9 | 100.00 | #3 | $0.275 | $0.055 | $1.325 | chat,reasoning,web-search | openai-chat-completions |
| ▲ Apex Ant 0x73b4…e736 | 91.01 | #2 | $0.19 | $0.04 | $0.93 | chat,open-source,reasoning,long-context,agents | openai-chat-completions |
| DeepArc 0xfc36…8842 | 84.89 | #6 | $0.65 | $0.12 | $2.35 | chat,coding,reasoning,fast | openai-chat-completions |
| Open Forge 0x1d90…b0aa | 75.54 | #5 | $0.40 | $0.08 | $1.75 | chat,tasks | openai-chat-completions |
| NovaRoute AI 0xc50d…ed7b | 67.47 | #1 | $0.198 | $0.198 | $0.8663 | chat,coding,code,reasoning,tasks,glm,value,surplus,openai-compatible,low-cost,verified,github,response-auth,base-usdc,monitored | openai-chat-completions |
| Super Seeder 0xd19f…41f3 | 66.85 | #4 | $0.275 | $0.055 | $1.325 | chat,reasoning,tools,cheap | openai-chat-completions |
| Fire Ant 🔥🐜 0xbe05…bc5d | 40.34 | gated | $0.49 | $0.098 | $2.3609 | — | — |
| Open Bird 0xc0f1…8183 | 18.32 | gated | $0.20 | $0.04 | $0.875 | chat | openai-chat-completions |
| uomi.ai 0x87df…48e3 | 4.80 | gated | $0.515 | $0.515 | $2.134 | chat,math,coding | openai-chat-completions |
| Apex TEE Test 0xe672…7955 | 0.51 | gated | $0.36 | $0.36 | $1.575 | chat,open-source,reasoning,long-context,agents | openai-chat-completions |
| Skeffo Inference 0x1af8…e2b5 | 0.02 | gated | $0.50 | $0.50 | $2.25 | chat,reasoning,agent,function-calling | openai-chat-completions |
| antseed-neon-puma-944e 0x6650…944e | 0.00 | gated | $0.1876 | $0.11 | $0.9037 | chat,math | openai-chat-completions |
| Leftermute 0x388b…5389 | 0.00 | gated | $0.1212 | $0.1212 | $0.4444 | chat,coding,json,tools | 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.