GLM-5vsGLM-4.6
Across 8 shared benchmarks, GLM-5 leads overall: GLM-5 wins 7, GLM-4.6 wins 0, with 1 ties and an average score difference of +16.69.
GLM-57 wins(88%)Ties1(0%)0 winsGLM-4.6
Benchmark scores
Grouped by capability, sorted by largest gap within each. 8 shared benchmarks.
Agent Level Benchmark
GLM-5 2/2| Benchmark | GLM-5 | GLM-4.6 | Diff |
|---|---|---|---|
| τ²-Bench - Telecom | 985 / 35 | 7127 / 35 | +27 |
| τ²-Bench | 89.704 / 40 | 75.9020 / 40 | +13.80 |
General Knowledge
GLM-5 2/2| Benchmark | GLM-5 | GLM-4.6 | Diff |
|---|---|---|---|
| HLE | 50.4018 / 157 | 30.4074 / 157 | +20 |
| GPQA Diamond | 8643 / 178Thinking (No Tools) | 82.9061 / 178 | +3.10 |
AI Agent - Information Search
GLM-5 1/1| Benchmark | GLM-5 | GLM-4.6 | Diff |
|---|---|---|---|
| BrowseComp | 75.9019 / 45 | 45.1038 / 45 | +30.80 |
Coding and Software Engineer
GLM-5 1/1| Benchmark | GLM-5 | GLM-4.6 | Diff |
|---|---|---|---|
| SWE-bench Verified | 77.8023 / 108Thinking (No Tools) | 6865 / 108 | +9.80 |
Instruction Following
GLM-5 1/1| Benchmark | GLM-5 | GLM-4.6 | Diff |
|---|---|---|---|
| IF Bench | 7210 / 29 | 4329 / 29 | +29 |
Math and Reasoning
Even 1/1| Benchmark | GLM-5 | GLM-4.6 | Diff |
|---|---|---|---|
| FrontierMath - Tier 4 | 2.1056 / 80Normal (No Tools) | 2.1056 / 80Normal (No Tools) | — |
Specs
| Field | GLM-5 | GLM-4.6 |
|---|---|---|
| Publisher | 智谱AI | 智谱AI |
| Release date | 2026-02-11 | 2025-09-30 |
| Model type | Chat model | Chat model |
| Architecture | MoE | MoE |
| Parameters | 744B | 355B |
| Context length | 200K | 200K |
| Max output | 128K | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | GLM-5 | GLM-4.6 |
|---|---|---|
| Text input | $1 / 1M tokens | Not public |
| Text output | $3.2 / 1M tokens | Not public |
| Cache write | $0.2 / 1M tokens | Not public |
One or both models have incomplete public pricing.
Summary
- GLM-5leads in:Agent Level Benchmark (2/2), General Knowledge (2/2), AI Agent - Information Search (1/1), Coding and Software Engineer (1/1), Instruction Following (1/1)
- Tied in:Math and Reasoning
On average across the 8 shared benchmarks, GLM-5 scores 16.69 higher.
Largest single-benchmark gap: BrowseComp — GLM-5 75.90 vs GLM-4.6 45.10 (+30.80).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.