Gemma 4 31B rose across 2 hosts, by up to 67% at DekaLLM (input)
Published 1 October 2026
Price move, per 1M tokens
- DekaLLM · input↑ 67%$0.060 → $0.100
- DekaLLM · outputunchanged$0.33 → $0.33
- DekaLLM · cache read— → $0.050
- DeepInfra through OpenRouter (fp8) · input↑ 15%$0.13 → $0.15
- DeepInfra through OpenRouter (fp8) · output↑ 5%$0.38 → $0.40
- DeepInfra's own listing (fp8) · input↑ 15%$0.13 → $0.15
- DeepInfra's own listing (fp8) · output↑ 5%$0.38 → $0.40
| host | rate | was | now | change |
|---|---|---|---|---|
| DekaLLM | input | $0.060 | $0.100 | ↑ 67% |
| DekaLLM | output | $0.33 | $0.33 | — |
| DekaLLM | cache read | — | $0.050 | — |
| DeepInfra through OpenRouter (fp8) | input | $0.13 | $0.15 | ↑ 15% |
| DeepInfra through OpenRouter (fp8) | output | $0.38 | $0.40 | ↑ 5% |
| DeepInfra's own listing (fp8) | input | $0.13 | $0.15 | ↑ 15% |
| DeepInfra's own listing (fp8) | output | $0.38 | $0.40 | ↑ 5% |
DekaLLM input: ↑ 67%DeepInfra through OpenRouter (fp8) input: ↑ 15%DeepInfra's own listing (fp8) input: ↑ 15%
If you buy
Budget for a higher input cost on this host before your next top-up, and re-check whether your workload's prompt-heavy mix still fits the spend you planned.
The model
Gemma 4 31Bopen the model →
Open weights31.3B parameters262K context18 hosts
intelligence48th of 168 writing50th of 168 coding49th of 168 agents55th of 55
Gemma 4 is a 31.3-billion-parameter text, image and video model from Google with a permissive Apache licence and a quarter-million-token request limit. Its measured coding skill outpaces its general text score, though it struggles on agent tasks and its speed varies sharply by provider.
Source
Read on openrouter.ai · we published this on 1 October 2026.
read the original at openrouter.ai ↗