Inference
EN
IBM

Granite 4.2 30B

Granite 4.2 30B is a reasoning model from IBM in the Granite 4.2 family. 17 benchmarks count toward its index, in 6 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2026-08
RankedEnough results for an index.
IndexAll categories, out of 100.41.6±5.4Median 51.0
CoverageIndex weight that has results.85%
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.128KMedian 262K
EloLMArena rating and rank.1361±11 #201Median 1422

Run locally

29.28B parameters
16-bit
62.5 GB
8-bit
33.7 GB
4-bit
19.6 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 24 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.4 counted42.0172 / 218
CodingCode writing and repair.4 counted43.2170 / 209
ReasoningLogic problems and puzzles.2 counted39.2168 / 227
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.4 counted35.0225 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted61.1—
MathMath problems.2 counted47.595 / 168

Results

17 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
American Invitational Mathematics Examination 2025Math89.2%18 / 53
Instruction Following BenchmarkInstructionAbove fitFar above this model's index.77.2%21 / 57
τ³-Bench Tool-Agent-User EvaluationAgentic62.0%15 / 32
Berkeley Function Calling Leaderboard v4Agentic61.4%13 / 26
Massive Multitask Language Understanding ProfessionalKnowledge77.6%43 / 75
LiveCodeBench v6Coding75.8%34 / 59
Artificial Analysis Humanity's Last ExamKnowledge11.2%138 / 222
Critical Physics TasksReasoning0.3%137 / 214
Scientific Code BenchmarkCoding38.8%23 / 35
Graduate-Level Google-Proof Q&AKnowledge66.4%99 / 141
Artificial Analysis GPQA DiamondKnowledge64.4%145 / 199
GDPval-AA normalizedAgentic3.7%97 / 133
Harvard-MIT Mathematics Tournament February 2025Math89.2%11 / 15
Artificial Analysis Long Context ReasoningReasoning49.0%160 / 217
Software Engineering Benchmark VerifiedCoding57.0%90 / 114
Artificial Analysis SciCodeCoding37.8%97 / 114
Terminal-Bench 2.1 · BenchLMAgentic29.2%81 / 86
SWE-bench ProCoding33.3%79 / 83
Artificial Analysis Omniscience AccuracyKnowledge11.2%160 / 164
Artificial Analysis Openness IndexKnowledge72.2%—

Sources

BenchLMCC BY-NC 4.0

History

5 releases
Index42.4 → 41.6
23 Oct 202612 Nov 2026

Questions

How good is Granite 4.2 30B?
Its index is 41.6 out of 100. That is place 202 of 283 ranked models.
What is Granite 4.2 30B best at?
Math: 47.5 out of 100, place 95 of 168.
What is the context window of Granite 4.2 30B?
128K tokens.
Is Granite 4.2 30B open source?
Yes. Its weights are published.
When was Granite 4.2 30B released?
August 2026.

Badge

Granite 4.2 30B on Inference360: place 202 of 283, index 41.6
HTML
<a href="https://inference360.ai/models/granite-4-2-30b"><img src="https://inference360.ai/badges/granite-4-2-30b.svg" alt="Granite 4.2 30B on Inference360: place 202 of 283, index 41.6" height="20"></a>
Markdown
[![Granite 4.2 30B on Inference360: place 202 of 283, index 41.6](https://inference360.ai/badges/granite-4-2-30b.svg)](https://inference360.ai/models/granite-4-2-30b)

Compare with

GPT-OSS 20B41.7GLM-4.641.8LFM3-1.7B41.5Claude 3.7 Sonnet Thinking41.4Granite 4.2 8B37.6

More from IBM

All models
Granite 4.2 8B37.6Granite 4.2 3B33.2Granite-4.0-H-1B16.3Granite-4.0-H-350M12.7Granite-4.0-350M11.9