Skip to content
Local AI Wizard
Your machine
Theme

granite-guardian-4.1-8b

ibm-granite · ibm-granite/granite-guardian-4.1-8b

Weights license: Apache 2.0 · code and components may use other terms

Reported by the publisher55 resultsibm-granite's model card · read 2026-10-25 · not ranked · see them
Memory needed~7 GBQ4 weights + working room, before the machine's reserve
Parameters8.4 BHub safetensors total
Repository updated2026-08-27A commit is not a release date

Quality

No comparable score

We track no public, comparable benchmark that covers this model yet. No score is inferred.

Reported by the publisher

55 results · ibm-granite's model card · read 2026-10-25
Aggregate F1
0.79granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
Aegis Safety Test
0.83granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
Beaver Tails
0.79granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
HarmBench Prompt
0.78granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
OAI hf
0.83granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
SafeRLHF test
0.79granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
Simple Safety Test
1.00granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
Toxic Chat
0.78granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
Every reported result55
xstest RH
0.90granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
xstest RR
0.44granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
xstest RR (h)
0.82granite-guardian-4.1-8b (non-think) · OOD Safety Benchmarks · thinking mode
Aggregate F1
0.78granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
Aegis Safety Test
0.85granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
Beaver Tails
0.77granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
HarmBench Prompt
0.76granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
OAI hf
0.82granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
SafeRLHF test
0.77granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
Simple Safety Test
1.00granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
Toxic Chat
0.74granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
xstest RH
0.86granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
xstest RR
0.45granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
xstest RR (h)
0.81granite-guardian-4.1-8b (think) · OOD Safety Benchmarks · thinking mode
AVG
0.760granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
AggreFact-CNN
0.598granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
AggreFact-XSum
0.763granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
ClaimVerify
0.757granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
ExpertQA
0.603granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
FactCheck-GPT
0.749granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
LFQA
0.883granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
RAGTruth
0.834granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
Reveal
0.889granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
TofuEval-MediaS
0.737granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
TofuEval-MeetB
0.767granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
Wice
0.783granite-guardian-4.1-8b (non-think) · RAG Hallucination Benchmarks · thinking mode
AVG
0.764granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
AggreFact-CNN
0.606granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
AggreFact-XSum
0.765granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
ClaimVerify
0.773granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
ExpertQA
0.605granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
FactCheck-GPT
0.752granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
LFQA
0.885granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
RAGTruth
0.841granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
Reveal
0.888granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
TofuEval-MediaS
0.730granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
TofuEval-MeetB
0.795granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
Wice
0.767granite-guardian-4.1-8b (think) · RAG Hallucination Benchmarks · thinking mode
fc_reward_bench
0.79granite-guardian-4.1-8b (non-think) · Function Calling Hallucination Benchmarks · thinking mode
fc_reward_bench
0.78granite-guardian-4.1-8b (think) · Function Calling Hallucination Benchmarks · thinking mode
GSM8k
93.71Granite Guardian 4.1 8B · Best-of-N Selection with Guardian as a Reward Model (JETTS)
MATH
50.79Granite Guardian 4.1 8B · Best-of-N Selection with Guardian as a Reward Model (JETTS)
HumanEval+
80.08Granite Guardian 4.1 8B · Best-of-N Selection with Guardian as a Reward Model (JETTS)
MBPP+
70.63Granite Guardian 4.1 8B · Best-of-N Selection with Guardian as a Reward Model (JETTS)
BigCodeBench
43.70Granite Guardian 4.1 8B · Best-of-N Selection with Guardian as a Reward Model (JETTS)
IFEval
82.81Granite Guardian 4.1 8B · Best-of-N Selection with Guardian as a Reward Model (JETTS)
Overall
70.29Granite Guardian 4.1 8B · Best-of-N Selection with Guardian as a Reward Model (JETTS)

Values as printed on the card, with the publisher's own settings (card section: Evaluations › OOD Safety Benchmarks; Evaluations › RAG Hallucination Benchmarks). Not comparable across publishers and not used in any ranking. Model card

Runs on

Fast memory holds it 90

The planning estimate fits the memory the model can use quickly: graphics card VRAM, or unified memory within the macOS GPU share.

Show all 90

Within memory, outside the fast part 26

Within the machine's memory, outside its fast part: on a graphics card the layers that do not fit run on the CPU from PC RAM (the listed RAM, or 32 GB assumed); on a Mac, above the macOS default GPU share. Not a speed figure.

Show all 26

One memory tier short 3

The same machines with the next memory size up would hold it in fast memory.

Hardware profiles grouped by the Q4 planning estimate on each machine (dedicated GPUs reserve 2 GB; unified and system memory 25%, 2 to 8 GB): the fast memory holds it, it is within memory but outside the fast part (layers offloaded to PC RAM, or above the macOS GPU share), or the machine is one memory tier short. Not a measured fit or speed.

Measured speed

No published measurement for this checkpoint

Tokens per second are shown only when a source publishes them with hardware, build and setup. They are never estimated.

How to run

Runtime compatibility and tokens per second have not been measured for this entry. Published metadata and a memory estimate do not prove a successful local run.

Library in source: transformers

Choose my computer

Technical details

What the source and the catalog store for this checkpoint. Missing fields are left out, never zero.

Task
text-generation
Architecture
GraniteForCausalLM
Library in source
transformers
Weight license
Apache 2.0
Commercial use
Allowed Apache License 2.0 text (SPDX) · read 2026-10-18license text
Every stored field14 fields
Revision checked
ab01ccca5dcfb80246369a086a4a87a29198f5af
Architecture
GraniteForCausalLM
Library
transformers
Hub task tag
text-generation
Declared base models
ibm-granite/granite-4.1-8b
Distributions in this group
ibm-granite/granite-guardian-4.1-8b-GGUF
Formats
safetensors
Access
Open download
Publisher kind
Official
Tensor evidence
floating-tensor-elements · 8.38B · BF16checked 2026-09-26
Repository created
2026-04-16creation is not the model's release
Repository updated
2026-08-27a commit is not a release date
Hub downloads (30 days) · likes
54,804 · 51this repository only · read 2026-10-25
Catalog state
Hub config or weights found; no inventory benchmark link

Sources

Identity and published specifications come from the checkpoint page. Benchmark scores belong to the named release and do not measure speed on your computer.

Valid source snapshot: 2026-09-26. This source record has not received a complete technical review. Missing data remain unknown; a repository commit is not a model release date.