IBM model product
Granite 3.3 8B
3-8B-Base is a decoder-only language model with a 128K token context window.
Updated Aug 12, 2026. Default version: Granite 3.3 8B Instruct
LLMBoard score3.8Granite 3.3 8B Instruct
Coverage100%14 benchmark families
Context window128KTokens
Official input priceN/AOfficial price unavailable
Granite 3.3 8B Specifications
Technical details for the model's default version.
- Version
- Granite 3.3 8B Instruct
- Released
- Apr 16, 2025
- Knowledge cutoff
- Apr 1, 2024
- Parameters
- 8B
- Context window
- 128K
- Max output
- 8.2K
- Inputs
- text
- Outputs
- text
- Open weights
- No
- License
- Apache 2.0
Granite 3.3 8B Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Granite 3.3 8B Benchmark Results
Benchmark scores for Granite 3.3 8B Instruct.
| AlpacaEval 2.0 | 62.7% | 02 | 4 | 66.7% | C | |
| AttaQ | 88.5% | 02 | 3 | 50.0% | C | |
| PopQA | 26.2% | 02 | 3 | 50.0% | C | |
| TruthfulQA | 66.9% | 03 | 18 | 88.2% | C | |
| HumanEval+ | 86.1% | 04 | 10 | 66.7% | C | |
| BIG-Bench Hard | 69.1% | 14 | 21 | 35.0% | C | |
| HumanEval | 89.7% | 14 | 66 | 80.0% | C | |
| Arena Hard | 57.6% | 15 | 26 | 44.0% | C | |
| DROP | 59.4% | 25 | 30 | 17.2% | C | |
| AIME 2024 | 81.2% | 26 | 53 | 51.9% | C | |
| MATH-500 | 69.0% | 32 | 32 | 0.0% | C | |
| GSM8k | 80.9% | 38 | 48 | 21.3% | C | |
| IFEval | 74.8% | 59 | 65 | 9.4% | C | |
| MMLU | 65.5% | 91 | 100 | 9.1% | C |
Granite 3.3 8B Arena Results
Preference and agent-evaluation results for the default version.
| text | overall | 313 | 1149.7 | 3,090 | N/A | |
| text style control | overall | 314 | 1208.1 | 3,090 | N/A |
Granite 3.3 8B Runtime Performance
Provider-specific output speed and catalog latency for Granite 3.3 8B Instruct. Runtime does not affect the capability score.
| Replicate | 50 tok/s | 0.3 s | 128K | 8.2K |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Granite 3.3 8B Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- N/A
- Tracked offerings
- 0
No provider prices
The default version has no current input or output token prices.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Granite 3.3 8B Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
| Granite 3.3 8B Base | N/A | 8.2B | N/A | N/A | No | Apache 2.0 | |
| Granite 3.3 8B Instruct | 3.8 | 8B | 128K | 8.2K | No | Apache 2.0 |
What is Granite 3.3 8B?
Key information about Granite 3.3 8B and its available data.
3-8B-Base is a decoder-only language model with a 128K token context window. 1-8B-Base by adding support for Fill-in-the-Middle (FIM) using specialized tokens, enabling the model to generate content conditioned on both prefix and suffix. This makes it well-suited for code completion tasks
Data as of 2026-08-11.
Granite 3.3 8B vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to Granite 3.3 8B
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
IB
IBM Granite 4.0 Tiny
IBM
GO
Google
Gemini Diffusion
AN
Claude Sonnet 3
Anthropic
AM
Nova Micro
Amazon
OP
GPT-4.1-nano
OpenAI
AC
Qwen2.5 Coder 32B
Alibaba Cloud / Qwen Team
FAQ
Common questions about Granite 3.3 8B.
When was Granite 3.3 8B released?
Granite 3.3 8B's default version was released on Apr 16, 2025.
How much does Granite 3.3 8B cost?
No official standard PAYG price is currently available for Granite 3.3 8B.
Who created Granite 3.3 8B?
Granite 3.3 8B was created by IBM.
What is the context window for Granite 3.3 8B?
The default version has a 128K token context window.
Is Granite 3.3 8B open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer Granite 3.3 8B?
No provider offering is currently linked to the default version.
What models should I compare Granite 3.3 8B with?
Nearby ranked alternatives include Qwen2.5 Coder 32B, GPT-4.1-nano, Nova Micro.