# Gemma 4 31B-it FP8 vs Glm 4.5 Air Fp8

On provider list prices, Glm 4.5 Air Fp8 costs $0.20 per million input tokens against $0.39 for Gemma 4 31B-it FP8: 1.9x apart. Output is $1.10 against $0.97. On Allocate both bill at list plus the 7% transaction fee.

## Specifications

| | Gemma 4 31B-it FP8 | Glm 4.5 Air Fp8 |
| --- | --- | --- |
| Lab | Google | Zai Org |
| Access | Open weights | Open weights |
| Context window | 256K tokens | 128K tokens |
| List price, input | $0.39 / M tokens | $0.20 / M tokens |
| List price, output | $0.97 / M tokens | $1.10 / M tokens |
| Cached input | n/a | n/a |
| License | Apache 2.0 | MIT |
| Fine-tunable | Yes | Yes |

Specifications and provider list prices from the Allocate catalog, checked 2026-07-08. Billed price is list plus the 7% transaction fee.

## What the numbers say

Take 1,000,000 requests a month at 1,200 input and 350 output tokens each. That workload costs $625 a month on Glm 4.5 Air Fp8 and $807.50 on Gemma 4 31B-it FP8 at list: a gap of $182.50, or 1.3x.

Gemma 4 31B-it FP8 reads 256K tokens per request against 128K for Glm 4.5 Air Fp8, 2.0x the window. That decides which one can take whole documents without splitting them.

## Choose Gemma 4 31B-it FP8 for

- The longer context window (256K vs 128K tokens)
- Fine-tuning under a permissive license (Apache 2.0)

## Choose Glm 4.5 Air Fp8 for

- The lower list price ($0.20 in / $1.10 out per M tokens)
- Fine-tuning under a permissive license (MIT)

## Common questions

### Which is cheaper, Gemma 4 31B-it FP8 or Glm 4.5 Air Fp8?

Glm 4.5 Air Fp8, on this workload shape. At list prices it is $0.20/$1.10 per million tokens in and out against $0.39/$0.97 for Gemma 4 31B-it FP8. Billed on Allocate: $0.21/$1.18 against $0.42/$1.04, list plus 7%.

### Which has the bigger context window?

Gemma 4 31B-it FP8: 262,144 tokens (256K) against 131,072 (128K) for Glm 4.5 Air Fp8.

### Can I fine-tune Gemma 4 31B-it FP8 or Glm 4.5 Air Fp8?

Both publish open weights (Gemma 4 31B-it FP8: Apache 2.0; Glm 4.5 Air Fp8: MIT), so both can be fine-tuned. On Allocate the trained weights stay inside your boundary and belong to you.

---

[HTML page](https://allocate.network/compare/google-gemma-4-31b-it-vs-z-ai-glm-4-5-air-fp8) · [Gemma 4 31B-it FP8](https://allocate.network/models/google-gemma-4-31b-it.md) · [Glm 4.5 Air Fp8](https://allocate.network/models/z-ai-glm-4-5-air-fp8.md) · [Machine-readable catalog](https://allocate.network/catalog.json)