Find & Fit
Filter by capability, price, and what fits.
160 of 160 models · cross-provider price from models.dev · smallest-GPU-that-fits computed live.
| Cache | Smallest rig | Rent $/hr | Weights | ||||||
|---|---|---|---|---|---|---|---|---|---|
| deepseek-ai/deepseek-v4-pro | 1.6T | 0.80 | $0.696routing-run | 76 | 6.9× | 89% | 3× mi455x (1296GB) | — | open |
| zai-org/glm-5.3 | 753B | 0.75 | $0 | — | — | – | 2× mi355x (576GB) | — | open |
| deepseek-ai/deepseek-v4.1-flashnew | 763B | 0.75 | $0.280amd | 49 | 8.6× | 96% | 2× mi355x (576GB) | — | open |
| deepseek-ai/deepseek-v3.2-speciale | 685B | 0.74 | $1.68 | — | — | – | 2× mi325x (512GB) | — | open |
| deepseek-ai/deepseek-v3.2 | 685B | 0.74 | $0.295orcarouter | 49 | 6.3× | 63% | 2× mi325x (512GB) | — | open |
| deepseek-ai/deepseek-v3.2-exp | 685B | 0.74 | $0.320 | — | — | – | 2× mi325x (512GB) | — | open |
| inclusionai/ring-2.6-1t | 1.0T | 0.74 | $2.50requesty | 3 | 1.0× | 67% | 2× mi455x (864GB) | — | closed |
| bb1975/mimo-v2.5-pro | 1.0T | 0.74 | $0 | — | — | – | 2× mi455x (864GB) | — | open |
| inclusionai/ring-2.5-1t | 1.0T | 0.74 | — | — | — | – | 2× mi455x (864GB) | — | open |
| inclusionai/ring-1t | 1000B | 0.74 | $2.24 | — | — | – | 2× mi455x (864GB) | — | closed |
| avlp12/inkling-975b-alis-mlx-dynamic-3.7bpw | 947B | 0.73 | — | — | — | – | 2× mi455x (864GB) | — | open |
| thinkingmachines/inkling | 952B | 0.73 | $4.05edenai | 24 | 2.3× | 88% | 2× mi455x (864GB) | — | open |
| redhatai/inkling-fp8-block | 952B | 0.73 | — | — | — | – | 2× mi455x (864GB) | — | open |
| willfalco/glm-5.2-exl3-tr3-3.36bpw | 175B | 0.73 | — | — | — | – | b100 (180GB) | — | open |
| blackfrost-ai/glm-5.2-abliterated-reap-nu176-nvfp4 | 273B | 0.73 | — | — | — | – | mi325x (256GB) | — | open |
1–15 of 160 models
1 / 11
Fit status applies to open-weights models with known params (Q4_K_M, 8K context, 90% usable VRAM, 1–8 GPUs); closed models show “fit unknown”, never “doesn't fit”. Cross-provider price, providers, spread, and cache from models.dev. Live data.