glm-4.6

glm-4.6

Zhipu AI’s flagship language model supports complex agent tasks and high-precision code generation.
2025-09-30
LLM
Model capability: thinkingModel capability: function_call
Input:
$0.286/1M tokensstarting from
Output:
$1.142/1M tokensstarting from
Bulk order? Contact your manager for exclusive deals

API Overview

GLM-4.6 is a flagship language model launched by Zhipu AI, featuring “longer context and stronger coding and reasoning capabilities.” It’s a versatile large model primarily designed as a high-performance inference engine that supportscomplex agent tasks and high-precision code generation.

  • Doubled Context Length:The context window has been expanded from 128K to 200K tokens, enabling effortless handling of ultra-long documents and multi-turn complex tasks.
  • Leading Code Capabilities:Outperforms in real-world scenarios such as Claude Code and Kilo Code, delivering more refined front-end page generation.
  • Significantly Enhanced Reasoning:Supports tool calls, with logic reasoning and multi-step task processing capabilities comprehensively surpassing GLM-4.5.
  • More Powerful Agents:Integrates more seamlessly into search-based and tool-based agent frameworks, ensuring more reliable task execution.
  • Natural Writing Style:Adopts a style that closely aligns with human preferences, making role-playing and creative writing more immersive.
  • Overall Superior Performance:Outperforms GLM-4.5 in all eight public benchmarks, rivaling Claude Sonnet 4 and DeepSeek-V3.1-Terminus.

───────────────────────────────────────────────────────────────────

Core Capabilities

📚 200K Ultra-Long Context:Supports ultra-long text understanding and generation, easily handling complex scenarios such as legal and scientific research.

💻 Code Generation Expert:In real-world applications, front-end rendering is more precise, with benchmark scores like LCB significantly ahead.

Tool-Augmented Reasoning:Natively supports calling external tools during reasoning, enabling a closed-loop “think-do” process.

🧰 Agent-Friendly Architecture:Deeply adapted to agent frameworks, making task decomposition and tool scheduling more efficient.

✍️ Human-like Writing Style:The tone is natural and fluent, making role-playing and multi-style content generation more authentic.

🏆 Multi-dimensional Performance Leadership:Outperforms previous generations across eight major benchmarks, firmly placing it among the top-tier models both domestically and internationally.

───────────────────────────────────────────────────────────────────

Related Evaluations

“Claude Sonnet 4.5 vs. GLM-4.6: A Battle of Chinese and Foreign Large Models in Programming—Has the Winner Been Decided Yet?”



Playground

Log in to explore more features! Click to Log In

API Analytics

API Reference (2)

API DescriptionAPI EndpointRequest MethodStabilityParameter Description
Chat (Zhipu GLM Multimodal)
POST
Stable
View Details
Chat (Zhipu GLM Multimodal)
POST
Stable
View Details

API Pricing

$
ModelDescriptionContextOfficial Price302.AI Price

glm-4.6

Input length [0, 32k], output length [0, 0.2k]
200000

Input$0.286 / 1M tokens
Output$1.142 / 1M tokens

Input$0.286/ 1M tokens
Output$1.142/ 1M tokens
Original Price

glm-4.6

Input length [0, 32k], output length [0.2k+]
200000

Input$0.43 / 1M tokens
Output$2 / 1M tokens

Input$0.43/ 1M tokens
Output$2/ 1M tokens
Original Price

glm-4.6

Input length [32k, 200k]
200000

Input$0.572 / 1M tokens
Output$2.29 / 1M tokens

Input$0.572/ 1M tokens
Output$2.29/ 1M tokens
Original Price