August 25, 2026 — IBM released Granite 4.2, its first family of dense, decoder-only reasoning language models in 3B, 8B, and 30B sizes. Every weight is Apache 2.0, every model exposes a thinking / non-thinking switch, and the 8B and 30B checkpoints add agentic reinforcement learning trained inside real coding and search environments — not just benchmark math.
If you build agents on open weights, the practical question is not IBM's press release. It is whether Granite 4.2 earns a slot next to Qwen 3.8 and Nemotron on your laptop or VPC this week.
TL;DR — what people are asking
| Question | Answer |
|---|---|
| When did it ship? | August 25, 2026 — Hugging Face, Ollama, GitHub |
| Sizes? | 3B, 8B, 30B — all dense, same template |
| License? | Apache 2.0 |
| Thinking mode? | On/off plus low-effort thinking for easy prompts |
| Tool calling? | Native OpenAI function-calling format via vLLM/SGLang |
| Agentic RL? | 8B and 30B only — terminal, code edit, web search sandboxes |
| Context? | Trained toward 512K; check each model card for served limits |
| Best local size? | 8B for single-GPU agents; 3B for edge probes |
What IBM actually shipped
IBM's research announcement frames Granite 4.2 as reasoning-first enterprise agents: plan before acting, call tools with explicit rationale, and stay dense enough to run without MoE routing complexity.
The Hugging Face builder post documents five pre-training phases (~15T tokens from scratch), SFT on chain-of-thought and agent trajectories, then:
- Foundational RL — all three sizes; math, science, coding, tool calling
- Agentic RL — 8B and 30B only; real sandboxed environments
- RLHF alignment — standard preference tuning
IBM also highlights CodeAlchemy synthetic code (roughly 1 trillion tokens) and a mid-training step before long-context extension — unusual transparency for a vendor open-weight drop.
What this means for what you build or pay
On-prem agent stacks: Granite 4.2 is aimed at teams that want reasoning + tools under Apache 2.0 without routing prompts to a closed API. Pair it with OpenCode or Codex OSS mode patterns — point the harness at a local vLLM endpoint and keep keys off the wire.
Cost vs cloud: A 30B dense model is not free to serve, but it is predictable — no per-token surprise bill. For compliance-heavy workflows, that trade often beats frontier API spend; see go open source AI for Fortune 500 for the procurement framing IBM is selling into.
Eval before swap: Do not retire Qwen or Nemotron on marketing copy. Run your agent evals (Terminal-Bench slices, internal ticket bots, RAG tools) on 8B first — IBM's agentic RL targets exactly those multi-step failures.
How to run Granite 4.2 this week
Ollama (fastest smoke test):
ollama pull granite4.2:8b
ollama run granite4.2:8b "Explain MoE routing in two paragraphs."
vLLM (tool-calling endpoint):
vllm serve ibm-granite/granite-4.2-8b-instruct \
--enable-auto-tool-choice --tool-call-parser granite
Point Claude Code with open models or any OpenAI-compatible client at http://localhost:8000/v1.
Toggle thinking mode in the model template when you need planning-heavy tasks; use non-thinking for latency-sensitive chat.
Granite 4.2 vs nearby open models
| Model | Architecture | Reasoning switch | Agentic post-train | License |
|---|---|---|---|---|
| Granite 4.2 8B | Dense | Yes | Sandbox RL | Apache 2.0 |
| Qwen 3.8 27B | Dense | Via prompting | Community + vendor SFT | Apache 2.0 |
| Nemotron 3.5 Lightning 30B | MoE | Via harness | NVIDIA agent recipes | NVIDIA open license |
| GLM-5.3 | MoE | Thinking variants | Cyber + code focus | MIT (weights) |
Granite's pitch is enterprise density + signed weights + published RL stages — not raw leaderboard margin.
Honest limitations
- Served context may lag training — IBM trained toward 512K; verify each card's runtime window before stuffing 200K-token repos into one prompt.
- 30B is not a laptop model — plan GPU memory like any dense 30B; quantization helps but agentic tool loops add overhead.
- No independent SWE-bench sweep yet — treat IBM's enterprise task claims as directional until you run your harness.
- Thinking tokens cost latency — low-effort mode helps easy prompts; hard agent tasks still burn tokens like any reasoning model.
Update — September 10, 2026: IBM open-sourced another model outside the Granite lineage — the NASA-IBM Lunar Foundation Model, built for lunar science rather than general reasoning.
Related on explainx.ai
- NASA-IBM Lunar Foundation Model: Open-Source AI for Moon Science
- Qwen 3.8 Flash-Next 125B MoE release
- NVIDIA Nemotron 3.5 Lightning 30B open MoE
- Codex open-source models with Ollama OSS mode
- What are agent skills?
- Go open source AI — Fortune 500 guide
- How to run open models locally with OpenCode
- Evaluating prompts — measure quality
- Top 10 open-weight models for laptops
IBM Granite 4.2 weights, Ollama tags, and vLLM recipes are accurate as of August 26, 2026 — verify model cards on Hugging Face before production deployment.