Inference
EN
IBM

Granite-4.0-H-350M

Granite-4.0-H-350M is a standard model from IBM in the Granite 4.0 350M family. 11 benchmarks count toward its index, in 6 categories.

  • Open weights
  • Standard
  • Input: text
  • Released 2025-10
RankedEnough results for an index.
IndexAll categories, out of 100.12.7±6.9Median 51.0
CoverageIndex weight that has results.80%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.32KMedian 262K
EloLMArena rating and rank.—Median 1422

Run locally

0.34B parameters
16-bit
1.7 GB
8-bit
1.4 GB
4-bit
1.2 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 8 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.1 counted12.5—
CodingCode writing and repair.1 counted5.0—
ReasoningLogic problems and puzzles.2 counted22.1222 / 227
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.4 counted11.5265 / 268
MultilingualTasks in many languages.1 counted5.0—
InstructionTasks with strict prompt rules.2 counted5.262 / 63
MathMath problems.None counted——

Results

11 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
Critical Physics TasksReasoning0.0%150 / 214
Artificial Analysis Humanity's Last ExamKnowledge6.4%167 / 222
Multilingual Grade School MathMultilingual16.2%12 / 13
Artificial Analysis Long Context ReasoningReasoning0.0%204 / 217
τ²-Bench Tool-Agent-User EvaluationAgentic14.6%141 / 149
Massive Multitask Language UnderstandingKnowledge36.2%33 / 34
Instruction-Following EvalInstruction61.6%68 / 70
Graduate-Level Google-Proof Q&AKnowledge26.1%138 / 141
Massive Multitask Language Understanding ProfessionalKnowledge14.4%74 / 75
Artificial Analysis IFBenchInstruction17.6%145 / 146
Evaluating Large Language Models Trained on CodeCoding38.0%22 / 22
Artificial Analysis GPQA DiamondKnowledge25.7%199 / 199
BIG-Bench HardReasoning33.3%—

Sources

BenchLMCC BY-NC 4.0

History

5 releases
Index12.8 → 12.7
23 Oct 202612 Nov 2026

Questions

How good is Granite-4.0-H-350M?
Its index is 12.7 out of 100. That is place 282 of 283 ranked models.
What is Granite-4.0-H-350M best at?
Reasoning: 22.1 out of 100, place 222 of 227.
What is the context window of Granite-4.0-H-350M?
32K tokens.
Is Granite-4.0-H-350M open source?
Yes. Its weights are published.
When was Granite-4.0-H-350M released?
October 2025.

Badge

Granite-4.0-H-350M on Inference360: place 282 of 283, index 12.7
HTML
<a href="https://inference360.ai/models/granite-4-0-h-350m"><img src="https://inference360.ai/badges/granite-4-0-h-350m.svg" alt="Granite-4.0-H-350M on Inference360: place 282 of 283, index 12.7" height="20"></a>
Markdown
[![Granite-4.0-H-350M on Inference360: place 282 of 283, index 12.7](https://inference360.ai/badges/granite-4-0-h-350m.svg)](https://inference360.ai/models/granite-4-0-h-350m)

Compare with

Granite-4.0-350M11.9MiniCPM5-1B15.5Granite-4.0-H-1B16.3

More from IBM

All models
Granite-4.0-H-1B16.3Granite-4.0-350M11.9Granite-4.0-1BUnrankedGranite 4.2 30B41.6Granite 4.2 8B37.6