Decision-Modelle
Modelle für die Decisions-API. Sie liefern typisierte Antworten (eine Wahrscheinlichkeit, eine Option aus einer Liste oder eine Stufe auf einer Skala) statt Text, für Agentenschritte wie Tool-Routing, Freigabeprüfungen und Grounding-Checks. Verfügbar über POST /decisions.
Empfohlen
-
'%3e%3cpath%20fill='%23d9d9d9'%20d='m41.04961%2080.271324l1.8897629%200l0%202.3307114l-1.8897629%200z'%20fill-rule='evenodd'/%3e%3cpath%20fill='%23d9d9d9'%20d='m42.221855%2081.45145l1.8897629%200l0%202.3307037l-1.8897629%200z'%20fill-rule='evenodd'/%3e%3cg%20filter='url(%23shadowFilter-g31e21232314_0_33.1)'%3e%3cuse%20xlink:href='%23g31e21232314_0_33.1'%20transform='matrix(1.0%200.0%200.0%201.0%200.0%202.0)'/%3e%3c/g%3e%3cdefs%3e%3cfilter%20id='shadowFilter-g31e21232314_0_33.1'%20filterUnits='userSpaceOnUse'%3e%3cfeGaussianBlur%20in='SourceAlpha'%20stdDeviation='2.0'%20result='blur'/%3e%3cfeComponentTransfer%20in='blur'%20color-interpolation-filters='sRGB'%3e%3cfeFuncR%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncG%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncB%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncA%20type='linear'%20slope='0.0'%20intercept='0'/%3e%3c/feComponentTransfer%3e%3c/filter%3e%3c/defs%3e%3cg%20id='g31e21232314_0_33.1'%3e%3cpath%20fill='%23d9d9d9'%20d='m42.22417%2028.470434l0%2055.307083l-27.653543%20-55.307083z'%20fill-rule='evenodd'/%3e%3c/g%3e%3cg%20filter='url(%23shadowFilter-g31e21232314_0_33.2)'%3e%3cuse%20xlink:href='%23g31e21232314_0_33.2'%20transform='matrix(1.0%200.0%200.0%201.0%200.0%202.0)'/%3e%3c/g%3e%3cdefs%3e%3cfilter%20id='shadowFilter-g31e21232314_0_33.2'%20filterUnits='userSpaceOnUse'%3e%3cfeGaussianBlur%20in='SourceAlpha'%20stdDeviation='2.0'%20result='blur'/%3e%3cfeComponentTransfer%20in='blur'%20color-interpolation-filters='sRGB'%3e%3cfeFuncR%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncG%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncB%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncA%20type='linear'%20slope='0.0'%20intercept='0'/%3e%3c/feComponentTransfer%3e%3c/filter%3e%3c/defs%3e%3cg%20id='g31e21232314_0_33.2'%3e%3cpath%20fill='%23d9d9d9'%20d='m42.223038%2083.77752l21.729656%200l18.653545%20-70.385826l-25.574802%2013.461943z'%20fill-rule='evenodd'/%3e%3c/g%3e%3cg%20filter='url(%23shadowFilter-g31e21232314_0_33.3)'%3e%3cuse%20xlink:href='%23g31e21232314_0_33.3'%20transform='matrix(1.0%200.0%200.0%201.0%200.0%202.0)'/%3e%3c/g%3e%3cdefs%3e%3cfilter%20id='shadowFilter-g31e21232314_0_33.3'%20filterUnits='userSpaceOnUse'%3e%3cfeGaussianBlur%20in='SourceAlpha'%20stdDeviation='2.0'%20result='blur'/%3e%3cfeComponentTransfer%20in='blur'%20color-interpolation-filters='sRGB'%3e%3cfeFuncR%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncG%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncB%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncA%20type='linear'%20slope='0.0'%20intercept='0'/%3e%3c/feComponentTransfer%3e%3c/filter%3e%3c/defs%3e%3cg%20id='g31e21232314_0_33.3'%3e%3cpath%20fill='%23fdb515'%20d='m41.0477%2027.293962l0%2055.30709l-27.653542%20-55.30709z'%20fill-rule='evenodd'/%3e%3c/g%3e%3cg%20filter='url(%23shadowFilter-g31e21232314_0_33.4)'%3e%3cuse%20xlink:href='%23g31e21232314_0_33.4'%20transform='matrix(1.0%200.0%200.0%201.0%200.0%202.0)'/%3e%3c/g%3e%3cdefs%3e%3cfilter%20id='shadowFilter-g31e21232314_0_33.4'%20filterUnits='userSpaceOnUse'%3e%3cfeGaussianBlur%20in='SourceAlpha'%20stdDeviation='2.0'%20result='blur'/%3e%3cfeComponentTransfer%20in='blur'%20color-interpolation-filters='sRGB'%3e%3cfeFuncR%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncG%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncB%20type='linear'%20slope='0'%20intercept='0.0'/%3e%3cfeFuncA%20type='linear'%20slope='0.0'%20intercept='0'/%3e%3c/feComponentTransfer%3e%3c/filter%3e%3c/defs%3e%3cg%20id='g31e21232314_0_33.4'%3e%3cpath%20fill='%2330a2ff'%20d='m41.046566%2082.60105l21.72966%200l18.653545%20-70.385826l-25.574806%2013.461943z'%20fill-rule='evenodd'/%3e%3c/g%3e%3c/g%3e%3c/svg%3e "vLLM")
d3 Vorschau
d3-preview - Hosting
- T-Cloud Deutschland (EU)
-
DiffusionGemma 26B A4B Vorschau
diffusiongemma-26B- A4B- it- Preview - Kontext
- 256K
- Hosting
- T-Cloud Deutschland (EU)
Beide sind Vorschau-Modelle auf T-Cloud. DiffusionGemma akzeptiert Text und Bild und hat ein Kontextfenster von 256K.
2 von 2 Modellen
| Ausgabeformate | Token-für-Token-Ausgabe via Server-Sent Events. Geeignet für niedrige Latenz und Echtzeit-UI. | Funktions-/Tool-Aufrufe (OpenAI-kompatibel). Das Modell kann strukturierte Tool-Aufrufe zurückgeben. | Preis pro Million Prompt-Tokens, die aus dem Prefix-Cache bedient werden: 15 % des normalen Eingabepreises. Gilt nur für T-Cloud-gehostete Sprachmodelle. | Details | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Google DiffusionGemma 26B A4B | Gehostet auf T-Cloud Public — der souveränen Infrastruktur der Telekom in Deutschland, innerhalb der EU. | '%3e%3cpath%20d='M15.9229%20345.043C52.0939%20442.527%20145.929%20512%20256%20512C366.071%20512%20459.906%20442.527%20496.077%20345.043L256%20322.783L15.9229%20345.043Z'%20fill='%23FFDA44'/%3e%3cpath%20d='M256%200C145.929%200%2052.0939%2069.472%2015.9229%20166.957L256%20189.217L496.077%20166.956C459.906%2069.472%20366.071%200%20256%200Z'%20fill='black'/%3e%3cpath%20d='M15.923%20166.957C5.633%20194.69%200%20224.686%200%20256C0%20287.314%205.633%20317.31%2015.923%20345.043H496.078C506.368%20317.31%20512%20287.314%20512%20256C512%20224.686%20506.368%20194.69%20496.077%20166.957H15.923Z'%20fill='%23D80027'/%3e%3c/g%3e%3cdefs%3e%3cclipPath%20id='clip0_76_3108'%3e%3crect%20width='512'%20height='512'%20fill='white'/%3e%3c/clipPath%3e%3c/defs%3e%3c/svg%3e){kind=small} Deutschland (EU) | Text, Bild | Text | ✓ | ✓ | 256K | 0,00 € | 0,00 € | 0,00 € | Kostenlos in jedem aktiven Tarif | |
| ||||||||||||
| vLLM d3 | Gehostet auf T-Cloud Public — der souveränen Infrastruktur der Telekom in Deutschland, innerhalb der EU. | Deutschland (EU) | Text, Bild | Text | ✗ | ✗ | — | 0,00 € | 0,00 € | 0,00 € | Kostenlos in jedem aktiven Tarif | |
| ||||||||||||
| Keine Modelle entsprechen den Filtern. | ||||||||||||