Iminingwane yemodeli
Malunga
NVIDIA: Llama 3.3 Nemotron Super 49B V1.5 yi a chat model eyenziwe ngu NVIDIA. Ivuma kuze kube yi 131K izithonjana zesimo ngasinye sombuzo. Ihamba ngezindlela zangaphandle — ~360 i-token umlayezo ngamunye (50% yokuphawula ngaphezulu kwezindleko eziphezulu).
Sebenzisa nge-API
curl https://api.free.ai/v1/chat/ \
-H "Authorization: Bearer YOUR_KEY" \
-d '{"model":"nvidia/llama-3.3-nemotron-super-49b-v1.5"}'
Imibuzo ebuzwa kaningi
NVIDIA: Llama 3.3 Nemotron Super 49B V1.5 yi a chat model eyenziwe ngu NVIDIA. Ivuma kuze kube yi 131K izithonjana zesimo ngasinye sombuzo. Ihamba ngezindlela zangaphandle — ~360 i-token umlayezo ngamunye (50% yokuphawula ngaphezulu kwezindleko eziphezulu).
NVIDIA: Llama 3.3 Nemotron Super 49B V1.5 isebenza kahle ku-ukukhuluma ngokujwayelekile, ukukusiza ukubhala, ukucabanga, usizo lwekhodi, nokuhlelwa. Zama isibonelo sokubanjwa phezulu ukuze ubone indlela yayo.
Malunga 360 ama-token ngomlayezo ophakathi. $1 ithenga ama-token angama-750,000, ngakho-ke ngisho namamodeli akhokhelwe abiza amasenti ngakunye. Ama-akhawunti amahhala athola ama-token angama-10,000 wokubhalisa kanye ne-pool yansuku zonke.
Kuxhomekeka kumsebenzi. /incoko/uthelekiso/ ukuvumela ukuthi uthumele isiphakamiso esifanayo ku NVIDIA: Llama 3.3 Nemotron Super 49B V1.5 nanoma iyiphi enye imodeli ebhekene ne-side-by-side — uthelekiso yindlela ekhawulela kakhulu yokucabanga.
Yebo. Izingeniso ziyokho — Free.ai ayisebenzisi ilungelo nganoma yini oyenzayo.
Ama-131,072 ama-token.
Iphendula umsuka we-token-by-token ngaphakathi ~1 sekondi. Isikhathi sokubuyiselwa konke sixhomekeke kude futhi kubukhulu bemodeli — amamodeli amancane asakazeka ngokushesha, amamodeli esifunda ahamba ngokushesha nge-deepness.
Yebo. Abasebenzisi abangeniswe babona yonke ingxoxo ku /account/?tab=history. Ungabelana futhi ngekhophi elinkiwe ngayinye yenoma iyiphi ingxoxo nge-Share button.
Free.ai ayiqeqeshi amamodeli emidlalweni yakho. Amamodeli ahlala e-GPUs ethu. Amamodeli aphezulu ahamba ku-upstream provider for inference.
Yebo. POST ku /v1/chat/ ngemodeli "nvidia/llama-3.3-nemotron-super-49b-v1.5" kanye nomyalezo we-array. Ukusakazwa kwe-SSE kuxhaswe. Ubufakazi obuphelele: /api/.
NVIDIA: Llama 3.3 Nemotron Super 49B V1.5 iyimodeli ephakeme ehlinzekwa ngumhlinzeki ongaphandle, ngakho-ke ukuhoxiswa kwe-self-hosting akukhona. Free.ai ibonisa ukuthengwa kwe-token-based.
Ama-akhawunti amahhala athola ama-token we-10,000 abhalisiwe kanye ne-pool yansuku zonke. Uma lokhu kudlula, phezulu uqala ku-$ 1 (ama-token we-750K) - akukho ubhaliso odingekayo.