Chii chaitika
Black Forest Labs (BFL) yakaburitsa FLUX 3 Action, 7-bhiriyoni-parameter yakavhurika huremu 'World Action Model' yakagadzirirwa marobhoti. Iyo modhi inotora zvinoonekwa nekamera, marobhoti mamiriro, uye mirairo yemutauro chaiwo kugadzira zviito zvemuviri. BFL inoshuma 42.92% chiyero chekubudirira paNVIDIA's RoboLab-120 benchmark, ichipfuura mutungamiri wekare, NVIDIA's Cosmos3-Nano-Policy (16B paramita), ne6.1 muzana mapoinzi. Iko kuburitswa kunosanganisira huremu, kodhi, uye zvakanaka-tuning mabikirwo, achibvumira vanogadzira kuti vagadzirise modhi kune chaiwo marobhoti vachishandisa masisitimu seLeRobot.
Black Forest Labs, inozivikanwa neiyo FLUX mufananidzo uye vhidhiyo modhi, yakawedzera kuita marobhoti nekuburitswa kweFLUX 3 Action. Iyi 7-bhiriyoni-parameter modhi ndeye yakavhurika-huremu 'World Action Model' (WAM) inogadzirisa zvinopinda nekamera, masisitimu ehurongwa, uye mirairo yemavara kuti ibudise nhevedzano yezviito zvemuviri makumi matatu nembiri pamwe nekufungidzira kwekushanduka kwechiitiko.
Sekureva kweVentureBeat, BFL inoshuma kuti FLUX 3 Action inowana 42.92% yakazara budiriro mwero paNVIDIA's RoboLab-120 benchmark. Ichi chibodzwa chinodarika chekare chepamusoro chakavhurika modhi, NVIDIA's Cosmos3-Nano-Policy, ne6.1 muzana mapoinzi. Zvikuru, FLUX 3 Action inoshandisa chete mabhiriyoni manomwe eparamende kana ichienzaniswa neCosmos's 16 bhiriyoni, uye BFL inoti inomhanya ka1.43 nekukurumidza, kumisa muganho mutsva wePareto wekuita zvakanaka nekuita.
Kuburitswa kunosanganisira uremu hwemuenzaniso, kodhi, kugadzira-tuning mabikirwo, uye mienzaniso inogoneka. BFL inotaura kuti zvikwata zvinokwanisa kugadzirisa modhi vachishandisa zviratidzo zvakaunganidzwa kubva kumarobhoti avo, vachinyatsotaura kuenderana neSO-101 robhoti uye Hugging Face's LeRobot chimiro. Iyo kambani yakaratidzawo kugona kweiyo modhi munzvimbo dzisiri-robhoti, kusanganisira drones dzinobhururuka uye kutamba mutambo wevhidhiyo Doom pasina kufa mu-mutambo, kunyangwe izvi zvichionekwa sekuyedza kwekutanga.
BFL first previewed FLUX 3 Action in July when launching the broader FLUX 3 family, but it was initially restricted to selected research and commercial partners. The current release marks a shift toward broader accessibility, although specific commercial API pricing for the Action model has not yet been announced. The model builds on BFL's Self-Flow research, which aims to learn useful representations of motion and cause-and-effect from video data without relying on separate frozen representation models.
Kwakabva mashoko: venturebeat.com ↗
Nei zvichikosha
Kuburitswa uku kwakakosha nekuti kunopa compact, yakavhurika-huremu imwe nzira kune makuru evaridzi kana akavharika marobhoti modhi. Nekuwana zvibodzwa zvepamusoro zvine ma paramita mashoma (7B vs 16B) uye nekukasira kufungidzira, BFL inodzikisa chipingamupinyi chekupinda kwezvikwata zvirikuda kuendesa marobhoti epamberi pasina zviwanikwa zvekombuta. Iyo yakavhurika-huremu hunhu inobvumira kukwenenzverwa kwenzvimbo uye kuvanzika kwedata, izvo zvakakosha kune maindasitiri maapplication uko proprietary data haigone kubva panzvimbo. Iyo zvakare inosimbisa iyo 'World Action Model' maitiro, apo vhidhiyo inozivisa kudzora kwemuviri, zvichigona kuderedza kudiwa kwekuunganidzwa kwedata rakawanda robhoti.
The release of FLUX 3 Action is consequential for the robotics industry because it offers a high-performance, open-weight option that challenges larger, more resource-intensive models. By achieving superior benchmark results with fewer parameters, BFL demonstrates that efficiency and performance are not mutually exclusive, which is critical for edge deployment where compute resources are limited.
For developers, the open-weight nature of the model is a key differentiator. It allows teams to inspect, modify, and fine-tune the model within their own infrastructure, ensuring that proprietary robot demonstrations and operational data remain private. This is a significant advantage over hosted robotics services or proprietary models like Google's Gemini Robotics On-Device 2, which are limited to selected testers.
The model's architecture, which combines video with action generation, aligns with the growing trend of using generative models to understand physical dynamics. This approach may reduce the amount of robot-specific data required for training, as the model can leverage broad visual and physical representations learned from video. This could accelerate the development of robotic systems for tasks such as sorting, assembly, and navigation.
However, the practical impact depends on the final license terms and the ability of outside teams to reproduce the reported results. Robotics benchmarks are often fragmented, and direct comparisons between different model architectures (WAMs, VLAs, and action-reasoning models) are difficult. The success of FLUX 3 Action will ultimately be determined by its performance in real-world, diverse robotic environments rather than just simulation benchmarks.
Interactive Mechanism: Iyo Inonyatsoshanda
Ongorora ari pasi tekinoroji kuseri kwekusimudzira uku uchipindirana.
In AI, what are a model's "parameters"?
Zvekutarisa zvinotevera
Vagadziri vanofanirwa kutarisa kuwanikwa kwematemu ekupedzisira erezinesi rekutengesa, sezvo kuburitswa kwazvino kuri kwekutsvaga uye kuwana shamwari. Kuzvimiririra kuberekazve kweiyo bhenji mhedzisiro pane dzakasiyana Hardware kwakakosha, sezvo marobhoti mabhenji anowanzo kushaya kumira. Pamusoro pezvo, tarisa kubatanidzwa kweFLUX 3 Chiito mune yakakura maAPIs ekutengesa kana kuti BFL inowedzera yakavhurika-musimboti zano rekubatanidza mamwe marobhoti-chaiwo magumo.
The most immediate factor to watch is the publication of the final commercial license terms for FLUX 3 Action. While BFL has released the weights and code, the specific conditions for commercial use, redistribution, and modification are not yet fully detailed. This will determine how widely the model can be adopted in industrial settings.
Independent verification of the benchmark results is crucial. BFL's reported 42.92% success rate on RoboLab-120 is based on their own testing, and NVIDIA's public leaderboard had not been updated with these results at the time of the report. Reproduction by third-party research groups will be necessary to confirm the model's performance and generalizability.
Developers should also monitor the integration of FLUX 3 Action into existing robotics frameworks and hardware. BFL has highlighted compatibility with the SO-101 robot and LeRobot, but the model's performance on other common robotic platforms, such as those from NVIDIA or Ai2, remains to be seen. The ease of on diverse hardware will be a key determinant of its adoption.
Pakupedzisira, nzvimbo yemakwikwi mune yakavhurika-huremu marobhoti iri kubuda nekukurumidza. Nemhando dzakadai seAi2's MolmoAct 2 uye NVIDIA's GR00T N1.7 yatovepo, budiriro yeFLUX 3 Action ichaenderana nekugona kwayo kupa mabhenefiti akasiyana maererano nekuita, kugona, uye nyore kushandisa. Mwedzi mishoma inotevera ingangoona mimwe mienzaniso uye mabhenji sezvo nharaunda inoedza modhi mukushandisa kwakasiyana.