Granite 4.1 8B
Apr 2026
Granite 4.1 8BCompare inon
Benchmarks of Granite 4.1 8B26
BenchmarkByScoreRank
Overall1
Knowledge4
Science2
Coding4
Writing11
Documents1
Legacy · saturated or no longer reported3
No benchmark matches
—per task?
self-host onlyprice
—output speed?
131Kcontext
Rankingin non-thinking mode?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-5 Minihighwins 62%—$0.053
OpenNVIDIA Nemotron 3.5 Lightning 30B A3B NVFP4wins 57%——Memory (4-bit): 4.4 GB → 11 GBBetterGPT-6 Lunamaxwins 69%—$0.068Function calling only with none · Chat Completions+3
BetterClaude Haiku 5.5highwins 70%—$0.079Remove sampling parameters+5BetterGPT-5.6 Lunaxhighwins 73%—$0.085ModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-5 Minihighwins 62%—$0.053GPT-6 Lunamaxwins 69%—$0.068Function calling only with none · Chat Completions+3Claude Haiku 5.5highwins 70%—$0.079Remove sampling parameters+5GPT-5.6 Lunaxhighwins 73%—$0.085GPT-5.6 Terraxhighwins 75%—$0.63Prompts with >272K input tokensModelScoreCostPer taskWhat changes
NVIDIA Nemotron 3.5 Lightning 30B A3B NVFP4Openwins 57%——Memory (4-bit): 4.4 GB → 11 GBGranite 4.2 8BunstatedOpenwins 55%—$0.06 / $0.25Context: 131K → 128K
Gemma 3 12BunstatedOpenwins 54%—$0.05 / $0.10Memory (4-bit): 4.4 GB → 6.1 GBGranite 4.2 3BunstatedOpensame—$0.03 / $0.12Context: 131K → 128K
Molmo2 8BunstatedOpensame——Context: 131K → 37KModelScoreCostPer taskWhat changes
Granite 4.2 30BunstatedOpenwins 61%—$0.16 / $0.65Context: 131K → 128K+1Granite 4.2 8BunstatedOpenwins 55%—$0.06 / $0.25Context: 131K → 128KGranite 4.2 3BunstatedOpensame—$0.03 / $0.12Context: 131K → 128KGranite 4.0 H SmallunstatedOpenwins 43%—$0.05 / $0.22Context: 131K → 128K+1Granite 4.1 30BOpenn/a—$0.10 / $0.25Memory (4-bit): 4.4 GB → 17 GBGranite 4.1 3BOpenn/a——Granite 4.0 350Mnon-thinkingOpenn/a——Context: 131K → 33KGranite 4.0 H 350MOpenn/a——Context: 131K → 33KAll IBM models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · compressed✓ 16 GB · 8-bit✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceApache-2.0free for commercial use
History
2 versionsBenchmarks
all categoriesChat & Knowledge
Humanity's Last Exam3.8#318
AA Omniscience Index-64.1#282
CritPt (physics)0#182
AA Intelligence Index 4.3.2≈6.6#287
AA Omniscience · Accuracy12.3#267
AA-LCR (long context) 1.113.3#256
GPQA Diamond43.3#272
Coding
Terminal-Bench 2.13.4#143
Agents & Tool Use
τ²-bench · AA27.8#157
Writing & Language
IFBench (instructions)38.6#166
FAQ
How much does Granite 4.1 8B cost?›
There is no per-unit API price: self-host only.
Is Granite 4.1 8B free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Granite 4.1 8B, and what is it best at?›
#315 of 336 in Chat & Knowledge, #95 of 97 in Coding, #155 of 242 in Agents & Tool Use, #151 of 197 in Writing & Language.
Is Granite 4.1 8B better than Molmo2 8B?›
On Arena Text · instruction following, Granite 4.1 8B scores 1,268.1 vs 1,270.2 — about the same.
What is the context window of Granite 4.1 8B?›
131K tokens.
When was Granite 4.1 8B released? Is there a newer version?›
Released 2026-04-29. It has been superseded by Granite 4.2 8B.
Is Granite 4.1 8B open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Granite 4.1 8B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Granite 4.1 8B read images?›
No, it accepts text only.
What is Granite 4.1 8B?›
Granite 4.1 8B is an open-weights language model from IBM, released Apr 2026.
What are the best alternatives to Granite 4.1 8B?›
GPT-5 Mini (better), NVIDIA Nemotron 3.5 Lightning 30B A3B NVFP4 (open weights).