GLOBALATLAS AI · DEDICATED INFERENCE
Habayn AI oo diyaarsan
Dooro habayn Qwen, DeepSeek ama Mistral oo leh checkpoint sax ah, qiimeyn xusuus GPU iyo dejinta serving-ka.
Laga bilaabo model ilaa habayntiisa
Dooro nooc model oo sax ah. Dhisme kasta wuxuu la imanayaa qiyaasta xusuusta inference-ka, dejinta serving-ka iyo hardware-ka katalogga ee iswaafaqaya. Helitaanka waxaa la hubiyaa ka dib xulashada; rakibidda waxaa la xaqiijiyaa ka hor dalbashada.
Qwen2.5-7B-Instruct
Kaaliyeyaasha qoraalka gaarka ah iyo chat-ka luqado badan
Model qoraal oo instructions-ka lagu habeeyey. Raadinta dukumentiyada waxay u baahan tahay adeeg retrieval/indexing ah oo gaar ah; sawirro iyo maqal kuma jiraan checkpoint-kan.
- GPU VRAM
- Assessment required
- RAM / disk la qorsheeyey
- 32 / 100 GB
- Context
- 8192 · BF16 · 1 codsi
Qwen2.5-14B-Instruct
Kaaliyeyaasha ganacsiga iyo habaynta dukumentiyada qaabaysan
Model qoraal oo instructions-ka lagu habeeyey. Raadinta dukumentiyada waxay u baahan tahay adeeg retrieval/indexing ah oo gaar ah; sawirro iyo maqal kuma jiraan checkpoint-kan.
- GPU VRAM
- Assessment required
- RAM / disk la qorsheeyey
- 64 / 150 GB
- Context
- 8192 · BF16 · 1 codsi
Qwen2.5-32B-Instruct
Tilmaamo adag iyo model luqadeed oo weyn oo iskiis loo marti-geliyo
Model qoraal oo instructions-ka lagu habeeyey. Raadinta dukumentiyada waxay u baahan tahay adeeg retrieval/indexing ah oo gaar ah; sawirro iyo maqal kuma jiraan checkpoint-kan.
- GPU VRAM
- Assessment required
- RAM / disk la qorsheeyey
- 128 / 200 GB
- Context
- 8192 · BF16 · 1 codsi
Mistral-7B-Instruct-v0.3
Adeegyada qoraalka iyo backend-yada codsiyada ee kooban
Model qoraal oo instructions-ka lagu habeeyey oo leh tokenizer-ka v3. Function calling wuxuu u baahan yahay chat template iswaafaqsan iyo fulinta qalabka dhinaca application-ka.
- GPU VRAM
- Assessment required
- RAM / disk la qorsheeyey
- 32 / 100 GB
- Context
- 8192 · BF16 · 1 codsi
DeepSeek-R1-Distill-Qwen-14B
Tijaabooyin reasoning iyo kaaliyeyaasha gaar ahaan domain-ka
Model reasoning oo gaar ah oo DeepSeek ah, lagana soo miiray wax-soo-saarka DeepSeek-R1 oo ku salaysan asaas Qwen2.5-14B ah. Ma aha model-ka buuxa ee 671B DeepSeek-R1 ama xirmo ka kooban laba model.
- GPU VRAM
- Assessment required
- RAM / disk la qorsheeyey
- 64 / 150 GB
- Context
- 8192 · BF16 · 1 codsi
DeepSeek-R1-Distill-Qwen-32B
Culeysyada reasoning-ka ee waaweyn ee kaabayaal gaar ah ku yaal
Model reasoning oo gaar ah oo DeepSeek ah, lagana soo miiray wax-soo-saarka DeepSeek-R1 oo ku salaysan asaas Qwen2.5-32B ah. Ma aha model-ka buuxa ee 671B DeepSeek-R1 ama xirmo ka kooban laba model.
- GPU VRAM
- Assessment required
- RAM / disk la qorsheeyey
- 128 / 200 GB
- Context
- 8192 · BF16 · 1 codsi
Ma u baahan tahay model kale ama cluster?
Atlas wuxuu qiimeyn doonaa shuruudahaaga wuxuuna u gudbin doonaa habaynta adag khubaradayada.
Atlas Ai Navigator →