gpt-oss-20b
OpenAIgpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...
Entrée
$0.03
Sortie
$0.13
Blended
$0.06
Où acheter ce modèle
13 routes d'accès| Route | Entrée | Sortie | Blended | vs |
|---|---|---|---|---|
| ● Darkbloom le moins cherfp8 | $0.02 | $0.10 | $0.04 | - |
| ● AkashML fp4 | $0.02 | $0.10 | $0.04 | +0% |
| ● CoreWeave fp4 | $0.03 | $0.13 | $0.06 | +50% |
| ● DekaLLM bf16 | $0.03 | $0.14 | $0.06 | +50% |
| $0.03 | $0.14 | $0.06 | +50% | |
| $0.03 | $0.15 | $0.06 | +50% | |
| ● Phala | $0.04 | $0.15 | $0.07 | +75% |
| $0.04 | $0.15 | $0.07 | +75% | |
| ● SiliconFlow fp8 | $0.04 | $0.18 | $0.08 | +100% |
| $0.05 | $0.20 | $0.09 | +125% | |
 Amazon Bedrock | $0.07 | $0.15 | $0.09 | +125% |
| $0.07 | $0.25 | $0.12 | +200% | |
| $0.07 | $0.30 | $0.13 | +225% |
● API officielle · ● plateforme cloud · ● hébergeur d'inférence · USD par million de tokens. Même modèle, fournisseurs différents - le prix et la quantization varient.
- Ctx
- 131K
- max output
- 118K
- cache read /1M
- $0.03
- Famille
- gpt
- Modalités
- text
- Fonctions
- json · reasoning · tools
Prix live, mis à jour le 2026-09-15.