Atlas

Models

← All models

Granite 4.1 30B

IBM · 2026-04-29 · 28.9B parameters

Granite 4.1 30B is IBM's 30B-class dense, decoder-only instruct language model, fine-tuned from Granite-4.1-30B-Base, with a 131,072-token context window. IBM trained the Granite 4.1 language models from scratch on approximately 15 trillion tokens, then applied supervised fine-tuning and a multistage reinforcement-learning pipeline to improve instruction following, chat, and tool calling without relying on extended reasoning traces. The weights are released under the Apache 2.0 license.

Benchmark scores

BenchmarkScore
AA-LCR18.7
AA-LCR v1.123.7
AA-Omniscience Index-67.8
Artificial Analysis Agentic Index1.3
Artificial Analysis Coding Index10.4
Artificial Analysis Intelligence Index8.9
Artificial Analysis Omniscience Accuracy13.9
Artificial Analysis Omniscience Hallucination Rate95.0
Artificial Analysis Openness Index61.1
Artificial Analysis Progress Index v1176
Capability123.4
CritPt (avg@5)0.0
GDPval-AA v2379
GDPval-AA v2 (normalized)0.0
GDPval-AA v2.1196
GPQA Diamond (Artificial Analysis avg@5)48.1
Humanity's Last Exam (Text-Only)4.1
IFBench (Artificial Analysis avg@5)44.4
SciCode (Artificial Analysis avg@3)25.8
Terminal-Bench 2.12.6
Terminal-Bench Hard2.3
τ²-bench (Telecom)42.1
𝜏³-Banking3.9
Loading Atlas data…