Granite 4.0 1B
Oct 2025
Granite 4.0 1BCompare inon
Benchmarks of Granite 4.0 1B15
BenchmarkByScoreRank
Overall1
Knowledge4
Science2
Coding1
Documents1
Legacy · saturated or no longer reported6
No benchmark matches
—per task?
self-host onlyprice
—output speed?
128Kcontext
Rankingin non-thinking mode?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-6 Lunalow×6.6—$0.005Function calling only with none · Chat Completions+3
OpenQwen3 VL 8B Instructnon-thinking×3.3—$0.18 / $0.69Memory (4-bit): 0.8 GB → 4.4 GBBetterGranite 4.2 3BthinkingOpen+50%—$0.006Memory (4-bit): 0.8 GB → 1.8 GB
BetterMinistral 3 3B 2512Open+46%—$0.008Memory (4-bit): 0.8 GB → 1.9 GB
BetterGemma 4 E4BthinkingOpen+40%—$0.008Memory (4-bit): 0.8 GB → 4 GBModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-6 Lunalow×6.6—$0.005Function calling only with none · Chat Completions+3Granite 4.2 3BthinkingOpen+50%—$0.006Memory (4-bit): 0.8 GB → 1.8 GBMinistral 3 3B 2512Open+46%—$0.008Memory (4-bit): 0.8 GB → 1.9 GBGemma 4 E4BthinkingOpen+40%—$0.008Memory (4-bit): 0.8 GB → 4 GBGPT-5.6 Lunalow×6.4—$0.01ModelScoreCostPer taskWhat changes
Qwen3 VL 8B Instructnon-thinkingOpen×3.3—$0.18 / $0.69Memory (4-bit): 0.8 GB → 4.4 GBQwen3 VL 8B (Reasoning)thinkingOpen×3.3—$0.18 / $2.09Memory (4-bit): 0.8 GB → 4.4 GBQwen 3.5 9BthinkingOpen×2.7—$0.21Memory (4-bit): 0.8 GB → 4.8 GBGemma 4 12BthinkingOpen×2.5—$0.10 / $0.30Memory (4-bit): 0.8 GB → 6 GBQwen3.5 4BthinkingOpen×2.5—$0.03 / $0.15Memory (4-bit): 0.8 GB → 2.3 GBModelScoreCostPer taskWhat changes
Granite 4.0 H Smallnon-thinkingOpen×2.3—$0.05 / $0.22Memory (4-bit): 0.8 GB → 19 GBGranite 4.1 30BOpen×2.3—$0.10 / $0.25Memory (4-bit): 0.8 GB → 17 GBGranite 4.1 8Bnon-thinkingOpen×2——Memory (4-bit): 0.8 GB → 4.4 GBGranite 4.2 8BthinkingOpen+82%—$0.024Memory (4-bit): 0.8 GB → 4.4 GBGranite 4.2 30BthinkingOpen+64%—$0.072Memory (4-bit): 0.8 GB → 18 GBGranite 4.0 Micronon-thinkingOpen+51%—$0.017 / $0.11Memory (4-bit): 0.8 GB → 1.6 GBGranite 4.1 3BOpen+50%——Memory (4-bit): 0.8 GB → 1.7 GBGranite 4.2 3BthinkingOpen+50%—$0.006Memory (4-bit): 0.8 GB → 1.8 GBAll IBM models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceApache-2.0free for commercial use
Benchmarks
all categoriesChat & Knowledge
Humanity's Last Exam4.8#284
AA Omniscience Index-81.6#318
CritPt (physics)0#180
AA Omniscience · Accuracy6.2#310
GPQA Diamond28.1#303
AIME 2025 · AA archived6.3#129
Coding
LiveCodeBench · AA archived4.7#149
Agents & Tool Use
τ²-bench · AA22.8#180
Writing & Language
IFBench (instructions)20.5#242
FAQ
How much does Granite 4.0 1B cost?›
There is no per-unit API price: self-host only.
Is Granite 4.0 1B free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Granite 4.0 1B, and what is it best at?›
#282 of 334 in Chat & Knowledge, #210 of 239 in Coding, #178 of 242 in Agents & Tool Use, #239 of 245 in Writing & Language.
Is Granite 4.0 1B better than Qwen3 0.6B?›
On AA Omniscience · Accuracy, Granite 4.0 1B scores 6.2 vs 6.3 — about the same.
What is the context window of Granite 4.0 1B?›
128K tokens.
When was Granite 4.0 1B released? Is there a newer version?›
Released 2025-10-28. It has been superseded by Granite 4.1 3B.
Is Granite 4.0 1B open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Granite 4.0 1B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Granite 4.0 1B read images?›
No, it accepts text only.
What is Granite 4.0 1B?›
Granite 4.0 1B is an open-weights language model from IBM, released Oct 2025.
What are the best alternatives to Granite 4.0 1B?›
GPT-6 Luna (better), Qwen3 VL 8B Instruct (open weights).