Njuolgga sisdollui

Flux 3 Action

Integrere

Modealla-id flux-3-action/so101
# Run the model; wait up to 20 s for the finished result in the same call
curl -X POST "https://vit.doitong.com/api/v1/run/flux-3-action/so101?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A red fox trots through fresh snow at sunrise, soft golden light, shallow depth of field",
    "scene_image_url": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg",
    "wrist_image_url": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg",
    "state": [
      -18.94,
      134.52,
      120.15,
      28.3,
      -32.17,
      27.02
    ]
  }'

# Still "queued" or "running"? Poll with the id from the response:
curl "https://vit.doitong.com/api/v1/requests/REQUEST_ID?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY"

Dát ohcamuš: ≈0,003 $ 3 API-kreditta Vuodje dán siiddu alde

Oza API-čoavdaga
Black Forest Labs Videos video #robotic#arm

Robota policy-modealla mii ovddida SO-101-gieđa doaibmabáttiid kamerarámmaid, jođus-stáhtusa ja instrukšuvnna vuođul

Black Forest Labs · Diehtu (JSON)

Dát lea du vuosttaš bohtos

Daga jearaldaga vuolábealde ja vuodje dan iežat API-čoavddain. Boađusat kopierejuvvojit min vurkemii, bisuhuvvojit álo dahje 7 beaivvi nuvttá.

Ráhkat jearaldaga

Kontrollat ráhkaduvvojit modealla skeama vuođul. Koda ja haddi ođasmahttojuvvojit go rievddadat daid.

•gáibiduvvo 88 / 2000

Robota bargodagaldagat čájehuvvon lunddolaš gielain.

•gáibiduvvo
Luoitte fiilla, dahje

Gurut tile-govva, mii čájeha fikserejuvvon scenekamera dálá rámma maŋŋá 256x256 rádjema.

•gáibiduvvo
Luoitte fiilla, dahje

Dálá gieđakamera rámma, olgeš tile:s ja 256x256 rádjejuvvon.

•gáibiduvvo

Mihtiduvvon dili árvvut: čieža pan ja lokten, čuolbma fleksan, gieđa fleksan ja rollen, buot grádaiguin, ja de grippera sajádat proseanttain.

Viiddiduvvon history_scene_image_url · history_wrist_image_url · num_inference_steps · guidance_scale · seed · state_history · command_history
Luoitte fiilla, dahje

Historjjálaš scenarámma mii gáibida `state_history`; guođđin geavaha dálá rámma. Sáddejuvvon rámma lea árimus historjjálaš observášuvdna, 7 čuoggá dás ovddabealde.

Luoitte fiilla, dahje

Gieđakamerarámma árimus historjjálaš čuoggás; `state_history` gáibiduvvo.

standárda 4

Euler-denoisinga lohku; checkpoint-valideren geavahedje 4 ceáhkká.

standárda 3

Classifier-free guidancea gieđahallandávda sihke video ja doaibma várás; bijo 1.0:a bisahit CFG eret.

Sampling-siedja DROID:a referánsaárvvuin: 0 ja SO-101:a: 42. Jus dan ii meroštuvvo, válljejuvvo siedja sattodaga.

Mihtiduvvon dili historjá; ođđa episode:s guođđojuvvo, muđui gáibida `command_history`. Bija mihtidemiid kronologalaš ortnegii 7 stivrenčuoggá ovddabealde `state`, 30 Hz.

Absoluhtta-kommandohistorjá mii gáibida `state_history`; maŋemus sáddojuvvon kommando lea maŋŋemus ráiddus. Ráidut čájehit kronologalaš 8 observašuvdnacuoggá ja meroštallet kommando juohkečuoggái.

POST /run/flux-3-action/so101?wait=20 haddemearrádus → 0,003 $
# Run the model; wait up to 20 s for the finished result in the same call
curl -X POST "https://vit.doitong.com/api/v1/run/flux-3-action/so101?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A red fox trots through fresh snow at sunrise, soft golden light, shallow depth of field",
    "scene_image_url": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg",
    "wrist_image_url": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg",
    "state": [
      -18.94,
      134.52,
      120.15,
      28.3,
      -32.17,
      27.02
    ]
  }'

# Still "queued" or "running"? Poll with the id from the response:
curl "https://vit.doitong.com/api/v1/requests/REQUEST_ID?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY"

Geavat dán ohcamuša

Du key báhcá dán fierpmádatlogan: dat sáddejuvvo dušše Doitong API:ii ja vurkejuvvo (localStorage) dušše jos válljet Remember. Viečča API-keya

Dát lea Flux 3 Action birra

Media sajis dát máhcaha JSON:a ovddiduvvon `actions`-ain, daid stivrenrate:in, čoahkkedáhtunamaiguin ja man ollu álggu dássaid čađahuvvojit (`execute_steps`) ovdal ođđa čuoldda bivdima. Dárbbašlaš sisavaldin lea teaksta `prompt` (gitta 2000 mearkka), `scene_image_url`, `wrist_image_url` ja 6-árvosaš `state`. Eaktodáhtolaš historjá-sisavaldin (`state_history`, `command_history`, historjá-govat) addá áigegova konteksta. `num_inference_steps` manná 1 to 8 ja `guidance_scale` 1 to 10.

Parametterat

NammaŠládja StandárdaDohkkehuvvon Govvádus
promptgáibiduvvo
Šládjastring
≤ 2 000 mearkka
Robota bargodagaldagat čájehuvvon lunddolaš gielain.
scene_image_urlgáibiduvvo
Šládjaurl · image
Gurut tile-govva, mii čájeha fikserejuvvon scenekamera dálá rámma maŋŋá 256x256 rádjema.
wrist_image_urlgáibiduvvo
Šládjaurl · image
Dálá gieđakamera rámma, olgeš tile:s ja 256x256 rádjejuvvon.
stategáibiduvvo
Šládjanumber[]
≤ 6 merki≥ 6 merki
Mihtiduvvon dili árvvut: čieža pan ja lokten, čuolbma fleksan, gieđa fleksan ja rollen, buot grádaiguin, ja de grippera sajádat proseanttain.
state_historyeaktodáhtolaš
Šládjaarray[]
≤ 7 merki≥ 7 merki
Mihtiduvvon dili historjá; ođđa episode:s guođđojuvvo, muđui gáibida `command_history`. Bija mihtidemiid kronologalaš ortnegii 7 stivrenčuoggá ovddabealde `state`, 30 Hz.
command_historyeaktodáhtolaš
Šládjaarray[]
≤ 8 merki≥ 8 merki
Absoluhtta-kommandohistorjá mii gáibida `state_history`; maŋemus sáddojuvvon kommando lea maŋŋemus ráiddus. Ráidut čájehit kronologalaš 8 observašuvdnacuoggá ja meroštallet kommando juohkečuoggái.
history_scene_image_urleaktodáhtolaš
Šládjaurl · image
Historjjálaš scenarámma mii gáibida `state_history`; guođđin geavaha dálá rámma. Sáddejuvvon rámma lea árimus historjjálaš observášuvdna, 7 čuoggá dás ovddabealde.
history_wrist_image_urleaktodáhtolaš
Šládjaurl · image
Gieđakamerarámma árimus historjjálaš čuoggás; `state_history` gáibiduvvo.
num_inference_stepseaktodáhtolaš
Šládjainteger
Standárda4
1 – 8
Euler-denoisinga lohku; checkpoint-valideren geavahedje 4 ceáhkká.
guidance_scaleeaktodáhtolaš
Šládjanumber
Standárda3
1 – 10
Classifier-free guidancea gieđahallandávda sihke video ja doaibma várás; bijo 1.0:a bisahit CFG eret.
seedeaktodáhtolaš
Šládjainteger | null
Sampling-siedja DROID:a referánsaárvvuin: 0 ja SO-101:a: 42. Jus dan ii meroštuvvo, válljejuvvo siedja sattodaga.
gáibiduvvo

Fiilaparametterat váldet almmolaš https:// URL:a, data: URI:a gitta 10 MB, dahje uploada file_url:a. Fiillat

Boađus

GET /requests/req_8k2m4q7x1z9c3v5b6n0p2r4t
{
  "id": "req_8k2m4q7x1z9c3v5b6n0p2r4t",
  "object": "request",
  "model": "flux-3-action/so101",
  "status": "succeeded",
  "created_at": "2026-10-20T12:00:00Z",
  "started_at": "2026-10-20T12:00:02Z",
  "completed_at": "2026-10-20T12:00:09Z",
  "queue_position": null,
  "cost": {
    "credits": 3,
    "usd": "0.003",
    "charged": 3,
    "refunded": 0
  },
  "input": {
    "prompt": "A red fox trots through fresh snow at sunrise, soft golden light, shallow depth of field",
    "scene_image_url": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg",
    "wrist_image_url": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg",
    "state": [
      -18.94,
      134.52,
      120.15,
      28.3,
      -32.17,
      27.02
    ]
  },
  "output": {
    "actions": [
      [
        1.0
      ]
    ],
    "action_labels": [
      "string"
    ],
    "action_hz": 1.0,
    "execute_steps": 1,
    "seed": 42,
    "timings": {}
  },
  "outputs": [
    {
      "type": "file",
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_size": 1843200,
      "stored": true
    }
  ],
  "error": null,
  "metrics": {
    "inference_seconds": 6.8
  },
  "urls": {
    "self": "https://vit.doitong.com/api/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t",
    "cancel": "https://vit.doitong.com/api/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t/cancel"
  }
}
Du fiillat, du válljenJuohke outputs[].url lea kopia min vurkemis: bisuhuvvo álo (mákso GB-mánu mielde) dahje 7 beaivvi nuvttá, du konto­o heiveheami dahje ?storage= mielde.
Seammá hápmi juohke modeallastype, url, content_type, width, height ja duration, govvii, videoii ja jienai seammá ládje. output lea modealla iežas boađusobjekta.
Outputa giettit
actions
array of array of numberOvddiduvvon doaibmaárra, mas kronologalaš ráiddut čájehit stivrenčuoggáid; árvvut eai leat ráddjejuvvon, nu ahte čuolbma-, leavggus- ja bargobáikarádjerádjut fertejit leat robotis. 30 Hz:s SO-101 addá 42x6 absoluhtta-kommanddosa grádaiguin ja grippera proseanttain; 15 Hz:s DROID addá 32x8 absoluhtta-čuolbmatárggiid radianain ja grippera gitta-dási proseanttain.
action_labels
array of stringNamahusat mat vástidit doaibmakolonnaid.
action_hz
numberDoaibmaráidduid stivrenfrekuensa.
execute_steps
integerÁlgodoaimmaid lohku maid čađat ovdal go bivddát ođđa oasi.
seed
integerSamplinga áigge geavahuvvon sattosiedja.
timings
objectServera áigemearrádusat sekunddain.

AI-agenttaide: dán siidu Markdowna lea dáppe https://doitong.com/se/developers/models/flux-3-action/so101.md. Buot modeallat ja ráđit: https://doitong.com/developers/llms.txt