token.app › Compare › GPT-5.4 vs GLM 5.2
GPT-5.4 vs GLM 5.2
GPT-5.4 leads on 5 of 6 shared benchmarks. GLM 5.2 is 15× cheaper per token on a blended 3:1 basis. GPT-5.4 accepts the larger context window (1.1M).
Head to head
Blended price uses a 3:1 input:output mix
| GPT-5.4 | GLM 5.2 | |
|---|---|---|
| Provider | OpenAI | Zhipu AI |
| Input $/1M | $2.50 | $0.252 |
| Output $/1M | $15.00 | $0.792 |
| Blended $/1M | $5.63 | $0.387 |
| Context window | 1.1M | 1.0M |
| Max output | 128K | 128K |
| Released | Mar 2026 | Jun 2026 |
| Licence | Proprietary | Open weights |
| GPQA Diamond | 93.3% | 91.9% |
| SWE-bench Verified | 76.9% | 78.7% |
| FrontierMath T1–3 | 78.6% | 59.2% |
| FrontierMath T4 | 49.0% | 29.3% |
| SimpleQA Verified | 44.8% | 38.1% |
| OTIS Mock AIME | 97.8% | 86.4% |
Benchmark scores independently run and published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0. Only benchmarks with a score for BOTH models are compared.
Choose GPT-5.4 if…
- Higher GPQA Diamond (93.3% vs 91.9%)
- Higher FrontierMath T1–3 (78.6% vs 59.2%)
- Higher FrontierMath T4 (49.0% vs 29.3%)
- Higher SimpleQA Verified (44.8% vs 38.1%)
- Larger context window (1.1M)
Choose GLM 5.2 if…
- Costs less per token — $0.387 vs $5.63 blended
- Higher SWE-bench Verified (78.7% vs 76.9%)
- Open weights — self-hostable
- Newer model (released Jun 2026)
FAQ
Which is better, GPT-5.4 or GLM 5.2?
Across the 6 benchmarks both models have independently-run scores for, GPT-5.4 leads on 5 and GLM 5.2 leads on 1. Benchmarks are one input — price and context window are on this page too.
Is GPT-5.4 cheaper than GLM 5.2?
GLM 5.2 is cheaper: $0.387 versus $5.63 per million tokens blended at 3:1 input:output. Input alone: $2.50 vs $0.252. Output alone: $15.00 vs $0.792.
What context windows do GPT-5.4 and GLM 5.2 have?
GPT-5.4: 1,050,000 tokens. GLM 5.2: 1,048,576 tokens.
Who makes GPT-5.4 and GLM 5.2?
GPT-5.4 is made by OpenAI. GLM 5.2 is made by Zhipu AI.