Z.ai

Z.ai is the consumer and developer surface for Zhipu AI, the Beijing research lab behind the open-weight GLM family of models. The free chat at chat.z.ai gives anyone a GLM-5.1-powered AI assistant, and the GLM Coding Plans plug an OpenAI-compatible endpoint into Claude Code, Cursor, Cline, Roo Code, OpenCode, and 15+ other AI IDEs at a fraction of what Anthropic or OpenAI charge for the same shape of usage.

What Z.ai does:

• •GLM-5.1 long-horizon coding the flagship model is tuned to run for hours on a single goal, sustaining context and self-correction across hundreds of iterative tool calls without losing the thread.
• •GLM-5-Turbo for fast, cheap inference a lower-latency, lower-cost model for code completion, light edits, and conversational tasks where deep reasoning is overkill.
• •OpenAI-compatible API drop the GLM endpoint into any client, IDE, or agent that speaks the OpenAI Chat Completions format, with no SDK rewrite or wrapper required.
• •20+ IDE integrations documented setup guides for Claude Code, Cursor, Cline, Roo Code, OpenCode, Aider, Continue, and more, with config snippets ready to paste.
• •MCP tools bundle Vision, Web Search, and Web Reader Model Context Protocol tools are included free on every paid plan, so coding agents can read screenshots, search the live web, and pull docs into context.
• •Open-source model weights GLM-5.1, GLM-5-Turbo, and earlier GLM checkpoints are published on HuggingFace under MIT, so teams can self-host on their own GPUs for air-gapped or compliance use cases.
• •Free chatbot tier chat.z.ai gives unauthenticated visitors a usable GLM-5.1 chat without sign-up for casual questions and prompt experiments.
• •Quota-based predictability coding plans use refresh-window quotas instead of metered billing, so a $9 plan can never become a $900 plan from one runaway agent loop.
• •Image generation endpoint a separate $0.015-per-image API exposes Zhipu's image model for builders who need integrated text-plus-image inside agentic workflows.
• •Active model release cadence Zhipu AI ships new GLM checkpoints and benchmark updates roughly monthly, with paid plans getting priority access on the release date.

Pricing:

• •Free chat $0/mo: chat.z.ai gives basic GLM access for casual chatbot use, with no MCP tools and no IDE plug-in.
• •GLM Coding Lite $9/mo (billed quarterly): roughly three times the usage of Claude Pro, full IDE integration, and 100 MCP tool calls per month.
• •GLM Coding Pro $27/mo: 5x the Lite quota, 40-60% faster inference, and 1,000 MCP tool calls per month for serious individual builders.
• •GLM Coding Max $72/mo: 4x the Pro quota, guaranteed peak-hour performance, 4,000 MCP tool calls per month, and first access to new GLM releases.
• •API tokens pay-as-you-go: GLM-5.1 at $1.40 per million input tokens and $4.40 per million output tokens, GLM-5-Turbo at $1.20 input and $4.00 output, image generation at $0.015 per image.

Limitations:

• •Coding plans are IDE-only and cannot power custom apps the quota plans are restricted to recognised coding tools, so SaaS apps and bespoke integrations must buy separate per-token API credits, which do not draw from the plan quota.
• •Quota refresh windows can interrupt long sessions when you exhaust the 5-hour rolling quota, the agent simply pauses with no overage option, which is great for cost control but can stall a long-running build mid-task.
• •Chinese-first product with international friction documentation, support, and payment flows are optimised for Chinese users, so Western teams report slower replies, occasional payment-method headaches, and clunkier onboarding compared with OpenAI or Anthropic.