Granite 4.0 1B

Granite 4.0 1B

IBM · Released Oct 28, 2025 Professional
Intelligence #10 / 576
82.0 Our Score
AA Index #362 / 378
7.3 Artificial Analysis
Input
— Not priced
Output
— Not priced
Context
— Not reported

Analysis Summary

Granite 4.0 1B is IBM's smallest parameter-count model in the Granite 4.0 family, with an intelligence index of 7.3 and a coding index of 2.9. Its most notable metric is the agentic index of 22.8, which is unusually high for a 1B model and likely reflects targeted tool-use training rather than broad capability.

In practice, MMLU Pro at 0.325 and livecodebench at 0.047 confirm that general reasoning and code generation are very limited. Instruction following (ifbench 0.205) and long-context reliability (lcr 0.04) are both weak. No pricing data is available.

Granite 4.0 1B is best understood as an on-device or embedded inference model for highly constrained environments. It is not suitable for client-facing content, coding assistance, or any workflow requiring reliable reasoning. Teams exploring ultra-small model deployment for specific classification or routing tasks may find it a useful baseline.

Assessed June 6, 2026

Editorial notes

Granite 4.0 1B is IBM's 1-billion-parameter model with a surprisingly high agentic index for its size, but very limited reasoning and coding capability make it unsuitable for substantive business tasks.

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?

Performance Profile

Intelligence1.4Technical1.2Value0Content1.4
Intelligence 1.4/10
Technical 1.2/10
Content 1.4/10
Value 0/10

How Granite 4.0 1B compares

Granite 4.0 1B ranks #362 of 378 AI models we track for overall intelligence, #292 of 315 for coding, #165 of 289 for agentic tasks. Granite 4.0 1B is currently free to use via OpenRouter.

Performance Indices

Source: Artificial Analysis

7.3 Intelligence Index
2.9 Coding Index
22.8 Agentic Index
6.3 Math Index

Benchmark Scores

Intelligence

GPQA Diamond 28.1% Graduate-level scientific reasoning
HLE 5.1% Humanity's Last Exam
MMLU Pro 32.5% Multi-task language understanding
AIME 2025 6.3% Competition mathematics (2025)
SciCode 8.7% Scientific computing

Technical

LiveCodeBench 4.7% Live coding evaluation
τ²-Bench 22.8% Conversational agent benchmark

Content

IFBench 20.5% Instruction following
LCR 4% Long-context reasoning

Benchmark data from Artificial Analysis and Hugging Face

How does Granite 4.0 1B stack up?

Compare side-by-side with other professional models.

Compare Models

Model Information

ProviderIBM
Release Date October 28, 2025
Status Active

Frequently asked questions about Granite 4.0 1B

How much does Granite 4.0 1B cost?

Granite 4.0 1B is currently available for free via OpenRouter.

Is Granite 4.0 1B good for coding?

On our coding benchmark index, Granite 4.0 1B ranks #292 of 315 models, placing it in the broader range of the field for code generation and debugging.

Who created Granite 4.0 1B?

Granite 4.0 1B is developed by IBM and was released on October 28, 2025.