glm-5.3TensorX Ltd.
- api
- chat
- hosting
- EU
- model lab
- Z.AI
- weights
- open
- added
- August 2026
- model id
- tensorx/glm-5.3
capabilities 5/8
6 providers serve glm-5.3. This page is one of them. Compare all endpoints
- input
- $1.75
- output
- $4.50
- cache write
- -
- cache read
- $0.44
worked cost
| volume | cost |
|---|---|
| 100K in + 10K out | $0.22 |
| 1M in + 100K out | $2.20 |
| 10M in + 1M out | $22.00 |
Requesty charges what the upstream provider charges, with no markup and no per-request fee. Prompt caching and smart routing cut the effective cost further on repeated context. Gateway pricing
- median ttft
- 917ms
- median tok/s
- 121.0
- blended /1M
- $0.65
- cache hit
- 87.5%
- answered
- 99.80%
Weekly aggregates from production traffic routed through Requesty to this endpoint. A metric with no qualifying sample is left out rather than filled in. Compare against the other providers
| eval | Score as a share of 100 | /100 |
|---|---|---|
| GPQA Diamond | 91.7 | |
| Coding Index | 74.8 | |
| SciCode | 59.0 | |
| Intelligence Index | 44.9 | |
| Humanity's Last Exam | 42.3 | |
| 5 evals | mean | 62.5 |
Scores from Artificial Analysis and public leaderboards, normalised to 100. Bar colour is the band, not the rank: green 80 and up, blue 55 and up, amber 30 and up. Benchmarks measure narrow skills, so test on your own workload before committing. Released 2026-08-18.
- retention
- none
- trains on prompts
- no
- hosted in
- EU
Terms are the provider's, not Requesty's: routing a request here puts it under them. TensorX Privacy Policy
Need every request to stay in one jurisdiction? EU hosting routes only through EU-hosted providers.
tensorx/glm-5.3this id pins the tensorx ltd. deployment, with no routing and no failover. the managed id on the canonical page routes across all 6 providers instead
from openai import OpenAI client = OpenAI( api_key="YOUR_REQUESTY_API_KEY", base_url="https://router.requesty.ai/v1",) response = client.chat.completions.create( model="tensorx/glm-5.3", messages=[ {"role": "user", "content": "Explain quantum computing in one paragraph."}, ],) print(response.choices[0].message.content)
Change the base url to https://router.requesty.ai/v1, use your Requesty key, set the model to the id above. Existing OpenAI SDK code needs no other edit, and the same key reaches every other model in the catalog. Browse all models
GLM 5.3 is the latest Z.ai flagship model for long horizon coding, reasoning, and agentic workflows. It builds on GLM 5.2 with a 1M token context window, stronger engineering task performance, and multiple thinking effort levels. Released under the MIT license, it is designed for sustained work over large repositories, complex tool use, and multi step technical tasks.
questions 6
How much does glm-5.3 cost?
What is the context window of glm-5.3?
How does glm-5.3 perform on benchmarks?
What can glm-5.3 do?
How do I use glm-5.3 with the OpenAI SDK?
Can I run glm-5.3 through Requesty?
more from tensorx ltd. 8
| endpoint | ctx | in /M |
|---|---|---|
| qwen3.8 | 1M | $2.50 |
| qwen3.8-2.4t-a95b | 1M | $2.50 |
| glm-5.3-flash | 1.0M | $0.20 |
| qwen3.8-flash-next | 262K | $0.20 |
| deepseek-v4-pro-0813 | 1.0M | $1.75 |
| deepseek-v4-flash-0731 | 1.0M | $0.25 |
| kimi-k3 | 1.0M | $3.00 |
| glm-5.2 | 1.0M | $1.50 |
call glm-5.3 through one endpoint
One key for this endpoint and 600+ other models. No markup on provider prices, automatic failover when a provider degrades, and prompt caching built in. Methodology