AI/TLDR

Granite 4.2 3B

Smallest member of IBM's Granite 4.2 family, released 25 August 2026: a 3B dense decoder-only reasoning model with a switchable thinking mode, published under Apache 2.0.

Overview

Granite 4.2 3B is the smallest of the three models IBM published on 25 August 2026 as its first family of dense, decoder-only reasoning language models. It carries the same explicit thinking mode as its larger siblings — chain-of-thought inside <think>…</think> tags before the answer, switchable off or down to a low-effort budget — in a footprint sized for on-premises and edge deployment. The weights are on Hugging Face under the Apache 2.0 license, with builds also distributed through Ollama and GitHub.

Training follows the family recipe: pre-training from scratch on roughly 15 trillion tokens across a five-phase schedule, supervised fine-tuning on about 7.2 million chain-of-thought, reasoning and agentic-trajectory samples, then a multi-stage GRPO reinforcement-learning pipeline. The 3B receives the foundational RL and RLHF alignment stages but not the agentic RL that the 8B and 30B run inside software-engineering, terminal and web-search environments, and IBM correspondingly leaves it unscored on the agentic coding benchmarks. Context is 128K tokens natively with a documented extension to 512K, tool calling uses the OpenAI-compatible function format, and twelve languages are supported.

On IBM's published evaluations the 3B scores 78.33 on AIME25, 69.71 on LiveCodeBench v6, 67.84 on MMLU-Pro, 54.80 on GPQA and 55.30 on RULER at 128K — competitive reasoning numbers for its size, with the gap to the 8B widest on conversational quality (Arena-Hard-V2 34.96 against 65.19). It runs under Transformers, vLLM or SGLang, with FP8, NVFP4, MXFP4 and GGUF quantizations published alongside the bf16 weights.

Released2026-08-25
LicenseApache-2.0
WeightsOpen weights
Parameters3B (dense)
Context128K (long-context extension to 512K)
ArchitectureDense decoder-only transformer — 40 layers, Grouped Query Attention (40 heads, 8 KV heads), RoPE, SwiGLU MLP, RMSNorm, bfloat16
ModalitiesText
StatusGenerally available — open weights on Hugging Face

Benchmarks

IBM's published Granite 4.2 evaluation table (25 August 2026), the three sizes side by side. A dash means the source left the cell blank — the 3B skips the agentic-RL benchmarks.

BenchmarkGranite 4.2 3BGranite 4.2 8BGranite 4.2 30B
AIME2578.33%86.67%89.17%
HMMT Feb2566.67%78.33%89.17%
GPQA54.8%64.14%66.41%
LiveCodeBench v669.71%73.24%75.77%
SciCode24.11%36.09%38.76%
SWE-bench Verified47.67%57%
SWE-bench Pro19.11%33.29%
SWE-bench Multilingual30.78%41.89%
Terminal-Bench 2.120.56%29.24%
τ³-bench45.78%58.06%62%
BFCL v452.41%50.29%61.39%
ProfBench32.1%41.2%42.9%
BirdBench41.07%41.85%
GDPval11891225
MMLU-Pro67.84%74.04%77.6%
Arena-Hard-V234.96%65.19%67.93%
IFBench (prompt)74.33%79.33%77.17%
RULER 64K67.52%80.99%89.96%
RULER 128K55.3%71.41%81.38%

Comparison source ↗

This model's scores

  1. AIME2578.33%
  2. IFBench (prompt)74.33%
  3. LiveCodeBench v669.71%
  4. MMLU-Pro67.84%
  5. RULER 64K67.52%
  6. HMMT Feb2566.67%
  7. RULER 128K55.3%
  8. GPQA54.8%
  9. BFCL v452.41%
  10. τ³-bench45.78%
  11. Arena-Hard-V234.96%
  12. ProfBench32.1%
  13. SciCode24.11%

Scores on a 0–100 scale (25-point gridlines); higher is better. Each benchmark links to its published source.

Strengths

  • Apache 2.0 weights — no licence restrictions on commercial use, fine-tuning or self-hosting
  • Switchable thinking mode in a 3B footprint: chain-of-thought, low-effort reasoning, or direct answers
  • Strong small-model reasoning — 78.33 on AIME25 and 69.71 on LiveCodeBench v6
  • 128K native context, documented to extend to 512K
  • Sized for on-premises and edge hardware, with GGUF and low-precision builds published at launch
  • 12-language support and native OpenAI-compatible tool calling

Best for

  • Edge and on-premises assistants where a 3B model is the largest that fits
  • Latency-sensitive tool-calling services that can run reasoning off or at low effort
  • Local development and prototyping through the GGUF and Ollama builds
  • Fine-tuning a small Apache-2.0 reasoning base for a narrow domain

Granite (open-weight) — every version

The full lineage of the Granite (open-weight) line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.

VersionReleasedContextLicense
Granite 4.2 30Bcurrent2026-08-25128KApache-2.0
Granite 4.2 8B2026-08-25128KApache-2.0
Granite 4.2 3B2026-08-25128KApache-2.0

FAQ

Is Granite 4.2 3B open source?

Yes. IBM publishes the weights on Hugging Face under the Apache 2.0 license, so commercial use, fine-tuning and self-hosting carry no licence restrictions. Builds are also distributed through Ollama and GitHub.

Can a 3B model really reason?

Granite 4.2 3B emits chain-of-thought inside <think>…</think> tags like its larger siblings, and on IBM's published evaluations it scores 78.33 on AIME25, 69.71 on LiveCodeBench v6 and 54.80 on GPQA. It trails the 8B on conversational quality — Arena-Hard-V2 34.96 against 65.19 — so the gap is widest on open-ended chat rather than on maths and code.

Why is it missing scores on SWE-bench and Terminal-Bench?

IBM ran the extra agentic reinforcement-learning stage — software-engineering, terminal and web-search environments — only on the 8B and 30B, and leaves the 3B unscored on those agentic coding benchmarks.

How long a context does it handle?

128K tokens natively, with a documented long-context extension to 512K. On IBM's RULER evaluation it scores 67.52 at 64K and 55.30 at 128K.