Granite 4.0 H 350M
Oct 2025
Granite 4.0 H 350MCompare inon
Benchmarks of Granite 4.0 H 350M13
BenchmarkByScoreRank
Overall1
Knowledge4
Science2
Coding1
Documents1
Legacy · saturated or no longer reported4
No benchmark matches
—per task?
self-host onlyprice
—output speed?
33Kcontext
Ranking?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterMinistral 3 3B 2512Open+53%—$0.008Memory (4-bit): 0.2 GB → 1.9 GB
OpenGemma 4 12Bthinking×4.2—$0.10 / $0.30Memory (4-bit): 0.2 GB → 6 GBBetterGemma 4 E4BthinkingOpen×2.5—$0.008Memory (4-bit): 0.2 GB → 4 GB
BetterGPT OSS 20BhighOpen×3.7—$0.011Memory (4-bit): 0.2 GB → 13 GBBetterMinistral 3 8B 2512Open+66%—$0.011Memory (4-bit): 0.2 GB → 4.5 GBModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
Ministral 3 3B 2512Open+53%—$0.008Memory (4-bit): 0.2 GB → 1.9 GBGemma 4 E4BthinkingOpen×2.5—$0.008Memory (4-bit): 0.2 GB → 4 GBGPT OSS 20BhighOpen×3.7—$0.011Memory (4-bit): 0.2 GB → 13 GBMinistral 3 8B 2512Open+66%—$0.011Memory (4-bit): 0.2 GB → 4.5 GB
Nemotron 3 Nano 30B A3BthinkingOpen×4.1—$0.017Memory (4-bit): 0.2 GB → 19 GBModelScoreCostPer taskWhat changes
Gemma 4 12BthinkingOpen×4.2—$0.10 / $0.30Memory (4-bit): 0.2 GB → 6 GB
Qwen 3.5 9BthinkingOpen×3.8—$0.21Memory (4-bit): 0.2 GB → 4.8 GBNVIDIA Nemotron 3 Nano 4BthinkingOpen×3.3——Memory (4-bit): 0.2 GB → 2 GB
LFM2.5 8B A1BthinkingOpen×3.2——Memory (4-bit): 0.2 GB → 4.2 GBTFalcon H1R 7BthinkingOpen×3.1——Memory (4-bit): 0.2 GB → 3.8 GBModelScoreCostPer taskWhat changes
Granite 4.1 30BOpen×2.5—$0.10 / $0.25Memory (4-bit): 0.2 GB → 17 GBGranite 4.1 8Bnon-thinkingOpen×2.2——Memory (4-bit): 0.2 GB → 4.4 GBGranite 4.1 3BOpen+92%——Memory (4-bit): 0.2 GB → 1.7 GBGranite 4.0 H Smallnon-thinkingOpen+79%—$0.05 / $0.22Memory (4-bit): 0.2 GB → 19 GBGranite 4.0 H 1BOpen+49%——Memory (4-bit): 0.2 GB → 0.7 GBGranite 4.0 Micronon-thinkingOpen+41%—$0.017 / $0.11Memory (4-bit): 0.2 GB → 1.6 GBGranite 4.0 1Bnon-thinkingOpen+17%——Memory (4-bit): 0.2 GB → 0.8 GBGranite 4.0 350Mnon-thinkingOpen-9%——All 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 Exam6.4#237
AA Omniscience Index-80.9#318
CritPt (physics)0#182
AA Intelligence Index 4.3.2≈4.8#329
AA Omniscience · Accuracy3.8#321
GPQA Diamond25.7#309
AIME 2025 · AA archived1.3#138
Coding
LiveCodeBench · AA archived1.9#152
Agents & Tool Use
τ²-bench · AA14.6#213
Writing & Language
IFBench (instructions)17.6#246
FAQ
How much does Granite 4.0 H 350M cost?›
There is no per-unit API price: self-host only.
Is Granite 4.0 H 350M free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Granite 4.0 H 350M, and what is it best at?›
#235 of 336 in Chat & Knowledge, #210 of 239 in Coding, #210 of 242 in Agents & Tool Use, #243 of 245 in Writing & Language.
Is Granite 4.0 H 350M better than LFM2 2.6B?›
On IFBench (instructions), Granite 4.0 H 350M scores 17.6 vs 19.5 — lower.
What is the context window of Granite 4.0 H 350M?›
33K tokens.
When was Granite 4.0 H 350M released? Is there a newer version?›
Released 2025-10-28. It is the current version in its line.
Is Granite 4.0 H 350M open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Granite 4.0 H 350M locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Granite 4.0 H 350M read images?›
No, it accepts text only.
What is Granite 4.0 H 350M?›
Granite 4.0 H 350M is an open-weights language model from IBM, released Oct 2025.
What are the best alternatives to Granite 4.0 H 350M?›
Ministral 3 3B 2512 (better), Gemma 4 12B (open weights).