vs
5:2
overall capability wins (0 ties across 7 scored). Granite 4.2 3B leads overall.
- Reasoning3:1
- Coding1:0
- Specs1:1
Verdict
Granite 4.2 3B leads coding (widest gap: +13.4 on SciCode); Granite 4.2 3B offers larger context (131K vs 131K); Granite 4.0 Micro allows ~16.0x more max output (131K vs 8K).
- Granite 4.2 3B leads coding (widest gap: +13.4 on SciCode)
- Granite 4.2 3B offers larger context (131K vs 131K)
- Granite 4.0 Micro allows ~16.0x more max output (131K vs 8K)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Granite 4.2 3B
- -
- Granite 4.0 Micro
- $0.13
Price per 1M tokens
Full comparison
Organization
IBM
IBM
Family
Granite
Granite
License
Apache 2.0
Apache 2.0
Open weights
Yes
Yes
Release
Aug 25, 2026
Oct 20, 2025
Knowledge cutoff
-
-
API / provider
No primary API price listed
IBM watsonx
Modalities
text → text
text → text
Specs
Context window
131K
131K
Max output
8K
131K
Parameters
3B
—
Pricing
Input $/1M
—
$0.06
Output $/1M
—
$0.27
Blended $/1M (3∶1)
-
$0.11
Speed
tok/s
-
130
TTFT (s)
-
0.15
Reasoning
MMLU-Pro
—
44.7
GPQA Diamond
55.9
33.6
Humanity's Last Exam
6.6
5.0
AIME 2025
—
6.0
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
9.2
9.3
AA-LCR v1.1
24.3
7.0
CritPt
—
—
MMMU-Pro
—
—
IFBench
—
24.8
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
18.0
Terminal-Bench 2.1
13.9
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
25.3
11.9
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
—
| Metric | Granite 4.2 3B | Granite 4.0 Micro | Delta |
|---|---|---|---|
| Identity | |||
| Organization | IBM | IBM | — |
| Family | Granite | Granite | — |
| License | Apache 2.0 | Apache 2.0 | — |
| Open weights | Yes | Yes | — |
| Release | Aug 25, 2026 | Oct 20, 2025 | — |
| Knowledge cutoff | - | - | — |
| API / provider | No primary API price listed | IBM watsonx | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 131K | 131K | A +72 |
| Max output | 8K | 131K | B +123K |
| Parameters | 3B | — | — |
| Pricing | |||
| Input $/1M | — | $0.06 | — |
| Output $/1M | — | $0.27 | — |
| Blended $/1M (3∶1) | - | $0.11 | — |
| Speed | |||
| tok/s | - | 130 | — |
| TTFT (s) | - | 0.15 | — |
| Reasoning | |||
| MMLU-Pro | — | 44.7 | — |
| GPQA Diamond | 55.9 | 33.6 | A +22.3 pts |
| Humanity's Last Exam | 6.6 | 5.0 | A +1.6 pts |
| AIME 2025 | — | 6.0 | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 9.2 | 9.3 | B +0.1 pts |
| AA-LCR v1.1 | 24.3 | 7.0 | A +17.3 pts |
| CritPt | — | — | — |
| MMMU-Pro | — | — | — |
| IFBench | — | 24.8 | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | 18.0 | — |
| Terminal-Bench 2.1 | 13.9 | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 25.3 | 11.9 | A +13.4 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | — | — |
Granite 4.2 3B
- Context
- 131K / 8K out
- Parameters
- 3B
- Price
- —
- Speed
- —
- Modalities
- text → text
- License
- Apache 2.0
Compact edge-oriented member of IBM's Granite 4.2 family (Apache 2.0, Aug 25, 2026) — a dense 3B reasoning model with the same native 128K context and switchable thinking mode as its 8B and 30B siblings.
Granite 4.0 Micro
- Context
- 131K / 131K out
- Parameters
- —
- Price
- $0.06 / $0.27
- Speed
- 130 tok/s · 0.15s TTFT
- Modalities
- text → text
- License
- Apache 2.0
Ultra-compact Granite 4 hybrid MoE for edge and high-volume watsonx pipelines — among the cheapest hosted enterprise LLMs.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.