3.1× less energy than GLM-5.
GLM-5 Environmental Impact
First frontier model trained entirely on Huawei Ascend — NVIDIA-free
The figures below are for one quick question. A longer chat with follow-ups uses about 0.77 Wh, and an hour or two of agent work about 22.0 Wh.
One query, in context
Energy, carbon and water for a single use. Quick single question (one exchange)
73% of a Google search.
About 1.0× a Google search and 6% of 10 m by car.
About 16× a drop of water and 16% of a teaspoon of water.
- Where it runsZhipu AI Cloud (China)
- Grid carbon intensity525 g CO₂/kWh · 33% renewable
- Compared with a Google searchabout the same as a search (0.3 Wh)
- Water~1,235 queries to fill 1 litre
Where GLM-5 sits
#61 of 199 by energy per query. Lighter than 70% of all models. Lit dots are the same kind of model; click any dot to open it.
Want something lighter?
1.5× less energy than GLM-5.
3.7× less energy than GLM-5.
What a month of use adds up to
Pick how much you use it. Based on one quick use at a time, five days a week.
- Energy
- 23.8 Wh
- Carbon
- 13.0 g
- Water
- 88 mL
A month of 5 chats a day, 5 days a week. The energy is about 2.0× a phone charge and 22% of a kettle boil.
Work out your own →- Energy
- 95.3 Wh
- Carbon
- 52.0 g
- Water
- 351 mL
A month of 20 chats a day, 5 days a week. The energy is about 7.9× a phone charge and 87% of a kettle boil.
Work out your own →- Energy
- 476.3 Wh
- Carbon
- 259.8 g
- Water
- 1.75 L
A month of 100 chats a day, 5 days a week. The energy is about 4.3× a kettle boil and 6% of a UK home's day.
Work out your own →How does GLM-5 compare?
Next to similar models. Bars use a log scale, so each step is a big jump.
Detailed breakdown
Energy
We estimate GLM-5 at ~0.22 Wh for a quick single question. CLEER benchmarked GLM-5.2 — the same 744B/40B-active architecture — directly at ~0.22 Wh on eight B300 accelerators; we carry that figure back to GLM-5. This is our extrapolation, not a CLEER figure or a measurement. It replaces our earlier ~1.2 Wh estimate, which was built on 2025 benchmark figures for older hardware.
Power source & carbon
Runs on Huawei Ascend hardware in Chinese data centres. The Ascend 910B's power efficiency differs from NVIDIA GPUs — exact comparisons are limited by lack of public data.
Water use
At ~0.81 mL per quick question, derived from the energy estimate at the ~3.7 L/kWh water factor we apply to this model's infrastructure. Water scales directly with energy, so longer chats, reasoning and agentic sessions use proportionally more.
About GLM-5
GLM-5 is an open-source text and chat model from Zhipu AI, released in January 20, 2026, that runs below the category average for energy consumption at 0.22 Wh per query. Because its weights are publicly available, it can be self-hosted on any infrastructure — meaning its carbon footprint depends entirely on where and how you choose to run it. At 744B parameters, it first frontier model trained entirely on huawei ascend — nvidia-free.
These figures are estimates derived from hardware specifications and API benchmarks — Zhipu AI has not published official energy data for GLM-5. Actual consumption may vary significantly depending on batching, quantisation, and infrastructure optimisations that we cannot observe from outside.
Key insights
- Open-source weights — can be self-hosted on infrastructure you control
The Zhipu AI GLM family
One chart per tier, on a shared time axis. Hover a dot for details; click to open that version.
What does your GLM-5 usage cost the planet?
Use the calculator to estimate your personal footprint based on how often you use GLM-5.
Frequently asked questions
How much energy does GLM-5 use per query?
Each GLM-5 query consumes approximately 0.22 Wh of energy. This is about the same as a traditional Google search (~0.3 Wh).
What is GLM-5's carbon footprint?
Based on the carbon intensity of Zhipu AI Cloud (China), each query produces approximately 0.12 g of CO2. The grid in this region has a carbon intensity of 525 g CO2/kWh with 33% renewable energy.
How much water does GLM-5 use?
Each query consumes approximately 0.81 mL of water, primarily used for cooling the data centers that process the request.
How does GLM-5 compare to a Google search?
A GLM-5 query uses about the same as a Google search in terms of energy. A Google search uses approximately 0.3 Wh, while GLM-5 uses 0.22 Wh.
Technical details
- Architecture
- MoE (40B active, trained on Huawei Ascend 910B)
- Parameters
- 744B
- Context window
- 128,000 tokens
- Release date
- 2026-01-20
- Open source
- Yes
- Training data cutoff
- 2025-12