Identity
- Status
- active
- Developer ID
ibm-granite/granite-4.2-3b- Released
- Aug 25, 2026
- Version
- Granite 4.2 3B
- License
- apache-2.0
IBMGranite 4.2 3B
Granite 4.2 3B is IBM's compact open-weight dense reasoning model with native 128K context, optional 512K long-context extension, flexible thinking modes, and tool calling.
- Token context
- 131K tokens
- Inputs
- Text
- Outputs
- Text
- Released
- Aug 25, 2026
Benchmark Market Position
Intelligence, Cost, and Efficiency
Rank #43 of 57
Claude Opus 5.5
GPT-5.5 Pro
Gemini 3.8 Flash- GPT-6.1 Sol
- GPT-6 Astra
Qwen3.8 Max
DeepSeek V4.1 Flash- Gemini 3.1 Pro
- GPT-5.6 Terra
Grok 4.7
ERNIE 5.1
Kimi K3
GLM 5.3- Qwen3.8 Flash
- Qwen3.7 Plus
- DeepSeek V4 Pro
- GLM 5.3 Flash
Mistral Large 4- Gemma 4 31B
- Qwen3.8 27B
- Qwen3.5 397B A17B
Hy3- Gemma 4 26B A4B
Inkling- Mistral Large 3
- GPT-6 Luna
NVIDIA Nemotron 3 Ultra 550B A55B BF16- Inkling Small
MiniMax M3- Kimi K2.7 Code
- GPT-5.4 Mini
- Gemini 3.5 Flash Lite
- Qwen3.5 35B A3B
- GPT-5.4 Nano
Muse Glimmer 30B
Solar Pro 4- DeepSeek R1
Olmo 3 32B ThinkNova 2 Lite
Command A- Granite 4.2 30B
- Granite 4.2 3B
- Granite 4.2 8B
- Llama 4 Maverick 17B 128E Instruct
- Llama 4 Scout 17B 16E Instruct
phi-4- NVIDIA Nemotron 3 Nano 30B A3B BF16
- NVIDIA Nemotron 3.5 Lightning 30B A3B
- GPT-4o
- GPT-4o Mini
- gpt-oss-120b
- gpt-oss-20b
Not ranked in current cohort
Not ranked in current cohort
The orange outline marks Granite 4.2 3B. Each available panel preserves its real rank and value; missing benchmark or pricing inputs remain explicitly unranked.
Providers & Pricing
Estimate workload cost →| Provider | Input / 1M | Output / 1M | Cache read / 1M | Cache write / 1M | Observed | Evidence |
|---|---|---|---|---|---|---|
 DeepinfraAvailable | $0.030 | $0.12 | $0.0075 | — | Oct 11, 2026 | Provider-reported |
Benchmarks
All benchmark results →| LMArena Text Arenatext-2026-10-15-a64ca3b12b6e | 1,292.94 | — | 3,251 | granite-4.2-3b; 95% CI [1281.43969143, 1304.43063080]; votes 3251; rank 253unknown | Original sourcewinner eligible | Third-party benchmark |
At a Glance
Model Facts
Capacity
- Token context
- 131K tokens
- Maximum output
- Unknown
- Total parameters
- 3.7B
- Active parameters
- Unknown
- Knowledge cutoff
- Not reported
Interface and Access
- Inputs
- Text
- Outputs
- Text
- API available
- Yes
- Open weights
- Yes
- Self-hostable
- Yes
- Reasoning
- Yes
- Vision
- No
- Tool calling
- Yes
- Capabilities
- chat, generation, reasoning, structured outputs, tools
This page represents one developer model product. Dated API snapshots, serving endpoints, pricing, regions, quantizations, and service tiers attach as versioned aliases or provider details and never create duplicate public model records.
Version Lineage
PredecessorUnknown
SuccessorsUnknown
Aliasesgranite-4.2-3b, Granite 4.2 3B, ibm-granite/granite-4.2-3b, ibm-granite/granite-4.2-3b-fp8, ibm-granite/granite-4.2-3b-GGUF, ibm-granite/granite-4.2-3b-mxfp4, ibm-granite/granite-4.2-3b-nvfp4
Explicit Unknowns
- Maximum output
- Tokenizer
- Knowledge cutoff
Primary Evidence
Sources and Observation Date
ibm-granite/granite-4.2-3b developer model repositoryIBM · observed Sep 2, 2026Deepinfra catalogDeepinfra · observed Oct 19, 2026ibm-granite/granite-4.2-3b config.json at e459acceac81IBM · observed Sep 2, 2026ibm-granite/granite-4.2-3b README.md at e459acceac81IBM · observed Sep 2, 2026ibm-granite/granite-4.2-3b immutable Hub API metadataIBM · observed Sep 2, 2026Introducing Granite 4.2IBM · observed Sep 2, 2026Hugging Face Publishers catalogHugging Face Publishers · observed Sep 2, 2026
Comparable Peers
Related Model Comparisons
| A | Pair | B | Context |
|---|---|---|---|
| vs | family variantstext | ||
| vs | cross-developer peerstext | ||
| vs | ERNIE 4.5 0.3B PTBaidu | cross-developer peerstext | |
| vs | Command A TranslateCohere | cross-developer peerstext | |
| vs | Gemma 4 12BGoogle DeepMind | cross-developer peerstext | |
| vs | Gemma 4 E2BGoogle DeepMind | cross-developer peerstext | |
| vs | family variantstext | ||
| vs | family variantstext | ||
| vs | family variantstext | ||
| vs | Phi-4 Mini InstructMicrosoft | cross-developer peerstext | |
| vs | Phi-4 Mini ReasoningMicrosoft | cross-developer peerstext | |
| vs | o3OpenAI | cross-developer peerstext |
Questions
Granite 4.2 3B FAQs
What is Granite 4.2 3B?+
Granite 4.2 3B is IBM's compact open-weight dense reasoning model with native 128K context, optional 512K long-context extension, flexible thinking modes, and tool calling. It is developed by IBM and its current sourced lifecycle status is active.
What inputs and outputs does Granite 4.2 3B support?+
Granite 4.2 3B's current record lists Text as input and Text as output.
How much context does Granite 4.2 3B support?+
The current record lists a 131K-token context window and does not report a maximum output length.
Is Granite 4.2 3B available through an API?+
Granite 4.2 3B is marked as API-available. The directly sourced provider records currently include Deepinfra.
How large is Granite 4.2 3B?+
The current record lists 3.7B total parameters and an unknown active parameter count. Its architecture is GraniteForCausalLM.
What capabilities and tasks does Granite 4.2 3B support?+
Granite 4.2 3B's recorded capabilities are chat, generation, reasoning, structured outputs, and tools. Its supported tasks are code, reasoning, and text.
How much does Granite 4.2 3B cost through an API?+
The lowest directly sourced prices currently attached to Granite 4.2 3B are $0.030 per million input tokens through Deepinfra and $0.12 per million output tokens through Deepinfra. Prices are provider-specific and should be checked against each cited observation date.
Can Granite 4.2 3B be self-hosted?+
Granite 4.2 3B is an open-weight model. Self-hosting is supported, and the recorded license is apache-2.0.
Does Granite 4.2 3B have related versions or aliases?+
No predecessor is recorded. No successor is recorded. Recorded aliases are granite-4.2-3b, Granite 4.2 3B, ibm-granite/granite-4.2-3b, ibm-granite/granite-4.2-3b-fp8, ibm-granite/granite-4.2-3b-GGUF, ibm-granite/granite-4.2-3b-mxfp4, and ibm-granite/granite-4.2-3b-nvfp4.
Which models can Granite 4.2 3B be compared with?+
The current compatible peer set includes Granite 4.2 30B, Olmo 3 7B Think, ERNIE 4.5 0.3B PT, Command A Translate, and Gemma 4 12B. Compatibility requires shared sourced modalities and tasks; it does not imply that one model is better. Browse model comparisons →