THUDM: GLM 4 32B

Text input Text output
Author's Description

GLM-4-32B-0414 is a 32B bilingual (Chinese-English) open-weight language model optimized for code generation, function calling, and agent-style tasks. Pretrained on 15T of high-quality and reasoning-heavy data, it was further refined using human preference alignment, rejection sampling, and reinforcement learning. The model excels in complex reasoning, artifact generation, and structured output tasks, achieving performance comparable to GPT-4o and DeepSeek-V3-0324 across several benchmarks.

Key Specifications
Cost
$$$
Context
32K
Parameters
32B
Released
Apr 17, 2025
Speed
Ability
Reliability
Supported Parameters

This model supports the following parameters:

Stop Presence Penalty Logit Bias Top P Temperature Seed Min P Frequency Penalty Max Tokens
Performance Summary

THUDM: GLM 4 32B, created on April 17, 2025, is a 32B bilingual model optimized for code generation, function calling, and agent-style tasks. It consistently performs among the fastest models and offers highly competitive pricing across all evaluated benchmarks. In terms of specific performance, the model demonstrates exceptional capability in Email Classification, achieving 99.0% accuracy, placing it in the 87th percentile. It also performs strongly in Reasoning tasks with 74.0% accuracy (74th percentile), indicating robust complex problem-solving abilities. General Knowledge is another area of strength, with 93.5% accuracy, though its percentile ranking is moderate. However, the model exhibits significant weaknesses in other critical areas. Its performance in Coding (Baseline) is notably poor, with only 1.0% accuracy (8th percentile), suggesting it is not suitable for this task despite its description. Similarly, Instruction Following yielded 0.0% accuracy, indicating a complete failure in this domain. Ethics (Baseline) also shows very low accuracy at 11.0% (11th percentile). While its speed is a consistent advantage, the model's high duration in Coding and Ethics benchmarks suggests inefficiency in these specific, challenging tasks. Overall, GLM 4 32B excels in classification and reasoning but struggles considerably with coding, instruction following, and ethical considerations.

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $0.24
Completion $0.24

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
Novita
Novita | thudm/glm-4-32b-0414 32K $0.24 / 1M tokens $0.24 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Free Executions Accuracy Cost Duration
Other Models by thudm