The small, fast IBM model is Granite 4.1 8B — the efficient tier at 118 tokens/sec and $0.050 per million input tokens. It trades a few points of raw intelligence for speed and cost, the right call for high-volume, latency-sensitive work. Granite 4.0 Micro (27 t/s) is next.
Compact, efficient models — the small/mini/flash/haiku tier — ranked by output speed. These trade a little raw intelligence for low cost and high throughput, which is the right tradeoff for chat, classification, extraction and other high-volume work.
The small, fast IBM model is Granite 4.1 8B — the efficient tier at 118 tokens/sec and $0.050 per million input tokens. It trades a few points of raw intelligence for speed and cost, the right call for high-volume, latency-sensitive work. Granite 4.0 Micro (27 t/s) is next.
Granite 4.0 Micro (27 t/s) is the closest alternative on this metric. See the full ranking above for the tradeoffs.
modelgrep tracks 2 IBM models with live benchmarks, speed, latency and per-provider pricing, led on intelligence by Granite 4.1 8B. 2 of them qualify for this ranking.