AI Benchmark Model
Granite 4.2 3B
Benchmark scores, pricing, speed, and model comparisons for Granite 4.2 3B.
Key scores
Granite 4.2 3B is an artificial intelligence model developed by IBM and released on August 25, 2026. This model features an intelligence index of 14.3 and a coding index of 17.5, while its math index remains unknown. Performance metrics for the model include an output speed of 117.258 tokens per second and a time to first token of 0.269 seconds. Benchmark results for the model are 0.559 on GPQA, 0.066 on HLE, 0.249 on SciCode, and 0.243333333333333 on LCR. Pricing for Granite 4.2 3B is set at $0.03 per million tokens for input and $0.12 per million tokens for output, resulting in a blended cost of $0.05 per million tokens.
Benchmark results
4 tracked benchmark rows for Granite 4.2 3B.
GPQA Reasoning
55.9% SciCode Other
24.9% LCR Other
24.3% HLE Other
6.6%
Comments (0)
to join the discussion
No comments yet
Be the first to share your thoughts!