Gemma 4 26B A4B
GoogleGemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...
Ent.
$0.09
Saí.
$0.30
Misto
$0.14
Onde comprar este modelo
11 rotas de acesso| Rota | Ent. | Saí. | Misto | vs |
|---|---|---|---|---|
| ● Darkbloom mais barato | $0.04 | $0.22 | $0.09 | - |
| ● DekaLLM bf16 | $0.06 | $0.33 | $0.13 | +44% |
| $0.07 | $0.34 | $0.14 | +56% | |
| ● NextBit bf16 | $0.09 | $0.30 | $0.14 | +56% |
| ● Cloudflare | $0.10 | $0.30 | $0.15 | +67% |
| ● Makora | $0.10 | $0.34 | $0.16 | +78% |
| ● Venice bf16 | $0.13 | $0.40 | $0.20 | +122% |
| $0.13 | $0.40 | $0.20 | +122% | |
| $0.13 | $0.40 | $0.20 | +122% | |
| ● SiliconFlow fp8 | $0.14 | $0.40 | $0.21 | +133% |
| $0.15 | $0.60 | $0.26 | +189% |
● API oficial · ● plataforma de nuvem · ● host de inferência · USD por 1M de tokens. Mesmo modelo, provedores diferentes — preço e quantização variam.
- Ctx
- 262K
- max output
- 236K
- cache read /1M
- $0.05
- Família
- gemma
- Modalidades
- image · text · video
- Recursos
- json · reasoning · tools
Preços ao vivo, atualizados 2026-09-15.