llms.blog
AI

GLM-5.3 Scores 60 on Artificial Analysis Intelligence Index, Matching Kimi K3

Independent AI evaluation platform Artificial Analysis has published its benchmark results for Z.ai's GLM-5.3, awarding the reasoning model a score of 60 on its Intelligence Index v4.1.1. The result places GLM-5.3 level with Moonshot AI's Kimi K3 and three points behind frontier leader Claude Opus 5 (63). The evaluation tested GLM-5.3 at its maximum reasoning effort configuration across a nine-part battery that measures agentic tool execution, terminal coding, graduate-level scientific problem-

axc
axc
2 min
GLM-5.3 Scores 60 on Artificial Analysis Intelligence Index, Matching Kimi K3

Independent AI evaluation platform Artificial Analysis has published its benchmark results for Z.ai's GLM-5.3, awarding the reasoning model a score of 60 on its Intelligence Index v4.1.1. The result places GLM-5.3 level with Moonshot AI's Kimi K3 and three points behind frontier leader Claude Opus 5 (63).

The evaluation tested GLM-5.3 at its maximum reasoning effort configuration across a nine-part battery that measures agentic tool execution, terminal coding, graduate-level scientific problem-solving, physics, factual reliability, and long-context comprehension. Across a comparison class of 181 models with a median score of 35, GLM-5.3 ranks eighth overall.

GLM-5.3 Evaluation on Artificial Analysis Intelligence Index

Post-Training Gains on Base Weights

Z.ai released GLM-5.3 on August 14, 2026, without initiating a new base pretraining run. The 753-billion parameter model uses the identical base weights as its predecessor, GLM-5.2. All measured performance improvements stem from scaled reinforcement learning applied across long-horizon task environments over a four-week post-training cycle.

Vendor-reported data highlighted significant jumps on specialized coding benchmarks, with Terminal-Bench 3.0 climbing from 4.6 to 28.3 and DeepSWE v1.1 rising from 46.2 to 66.9. Artificial Analysis's independent test run confirms that these post-training gains translate into broader general reasoning and tool-use capabilities under standardized third-party evaluation.

Inference Cost and Token Economics

While GLM-5.3 matches Kimi K3's score of 60, the two models differ significantly in API pricing and token consumption patterns:

  • Token Pricing: Z.ai lists GLM-5.3 at $1.40 per million input tokens and $4.40 per million output tokens. In comparison, Moonshot prices Kimi K3 at $3.00 per million input and $15.00 per million output tokens.
  • Cost Per Evaluation Task: Across the Intelligence Index benchmark, GLM-5.3 averaged $0.68 per task, compared to $0.84 for Kimi K3 and $2.34 for Claude Opus 5. The full evaluation run across the suite cost $1,238.50 on Z.ai's API.
  • Generation Verbosity: GLM-5.3 exhibited high output token volume during reasoning steps, generating 170 million output tokens across the entire evaluation battery against a class median of 72 million tokens.

Deployment and Planned Open Weights

GLM-5.3 is currently accessible through Z.ai's hosted API and GLM Coding Plan subscriptions. The model enforces active reasoning traces and supports three selectable effort tiers, rejecting API calls that attempt to disable reasoning mode.

Z.ai has scheduled an open-weights release two weeks after launch following final safety evaluations, which would establish GLM-5.3 alongside Kimi K3 as the highest-scoring open-weight architectures on the Artificial Analysis leaderboard.

Sources

AILLMBenchmarksReasoningEvals
ShareX (Twitter)LinkedIn

Written by

axc

More to read

  • Listwise Reranking in Production: Comparing Pointwise, Pairwise, and Listwise LLM Architectures, Sliding-Window Permutations, and Serving Economics

    Information retrieval systems in production Retrieval-Augmented Generation (RAG) and enterprise search have transitioned through multiple reranking paradigms. While first-stage retrieval (dense vector embeddings and sparse lexical BM25/SPLADE) retrieves candidate sets of 50 to 200 documents in under 20 milliseconds, the precision of downstream generation depends heavily on the reranking stage. Traditional neural rerankers evaluate candidates through pointwise scoring or pairwise classification.

    1 min

  • The Lottery Ticket Hypothesis in Large Language Models: How Sparse Subnetworks and Iterative Magnitude Pruning Retain Transformer Capacity

    The Lottery Ticket Hypothesis in Large Language Models: How Sparse Subnetworks and Iterative Magnitude Pruning Retain Transformer Capacity Modern large language models operate under extreme overparameterization. Frontier architectures allocate tens or hundreds of billions of parameters to achieve low perplexity and robust generalization across reasoning, code generation, and factual retrieval. Yet empirical pruning consistently demonstrates that post-training models can lose 30% to 50% of their

    1 min

  • Enterprises Curb AI Agent Autonomy Amid 40% Project Cancellation Projections

    Enterprises deploying agentic artificial intelligence are shifting architectures away from open-ended autonomy toward bounded, verifiable execution as projects encounter governance, security, and financial bottlenecks in production environments. Data from industry research firms highlights a widening divergence between model capability and operational control. According to projections from Gartner, more than 40% of current agentic AI initiatives are projected to be canceled by the end of 2027.

    1 min