Meta model product
Llama 3.2 3B
2 3B Instruct is a large language model that supports a context length of 128K tokens and are state-of-the-art in their class for on-device use cases like summarization, instruction following, and rewriting tasks running locally at the.
Updated Aug 12, 2026. Default version: Llama 3.2 3B Instruct
LLMBoard score0.0Llama 3.2 3B Instruct
Coverage80%10 benchmark families
Context window128KTokens
Official input priceN/AOfficial price unavailable
Llama 3.2 3B Specifications
Technical details for the model's default version.
- Version
- Llama 3.2 3B Instruct
- Released
- Sep 25, 2024
- Knowledge cutoff
- Unknown
- Parameters
- 3.2B
- Context window
- 128K
- Max output
- 128K
- Inputs
- text
- Outputs
- text
- Open weights
- No
- License
- Llama 3.2 Community License
Llama 3.2 3B Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Llama 3.2 3B Benchmark Results
Benchmark scores for Llama 3.2 3B Instruct.
| InfiniteBench/En.MC | 63.3% | 01 | 1 | 100.0% | C | |
| InfiniteBench/En.QA | 19.8% | 01 | 1 | 100.0% | C | |
| NIH/Multi-needle | 84.7% | 01 | 1 | 100.0% | C | |
| Open-rewrite | 40.1% | 01 | 1 | 100.0% | C | |
| TLDR9+ (test) | 19.0% | 01 | 1 | 100.0% | C | |
| BFCL v2 | 67.0% | 04 | 5 | 25.0% | C | |
| Nexus | 34.3% | 04 | 4 | 0.0% | C | |
| ARC-C | 78.6% | 17 | 34 | 51.5% | C | |
| HellaSwag | 69.8% | 24 | 27 | 11.5% | C | |
| MGSM | 58.2% | 27 | 31 | 13.3% | C | |
| GSM8k | 77.7% | 40 | 48 | 17.0% | C | |
| IFEval | 77.4% | 56 | 65 | 14.1% | C | |
| MATH | 48.0% | 60 | 71 | 15.7% | C | |
| MMLU | 63.4% | 96 | 100 | 4.0% | C | |
| GPQA | 32.8% | 219 | 234 | 6.4% | C |
Llama 3.2 3B Arena Results
Preference and agent-evaluation results for the default version.
| text | overall | 331 | 1109.7 | 7,936 | N/A | |
| text style control | overall | 339 | 1166.4 | 7,936 | N/A |
Llama 3.2 3B Runtime Performance
Provider-specific output speed and catalog latency for Llama 3.2 3B Instruct. Runtime does not affect the capability score.
| DeepInfra | 171.5 tok/s | 0.24 s | 128K | 128K |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Llama 3.2 3B Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $0.02 input, $0.02 output per 1M via Inference
- Tracked offerings
- 11
| Nvidia | meta/llama-3.2-3b-instruct | global | N/A | N/A | 32.8K | |
| Inference | meta/llama-3.2-3b-instruct | global | $0.02 | $0.02 | 16K | |
| LLM Gateway | llama-3.2-3b-instruct | global | $0.03 | $0.05 | 32.8K | |
| NovitaAI | meta-llama/llama-3.2-3b-instruct | global | $0.03 | $0.05 | 32.8K | |
| NanoGPT | meta-llama/llama-3.2-3b-instruct | global | $0.0306 | $0.0493 | 131.1K | |
| OpenRouter | meta-llama/llama-3.2-3b-instruct | global | $0.05 | $0.33 | 131.1K | |
| Cloudflare Workers AI | @cf/meta/llama-3.2-3b-instruct | global | $0.0509 | $0.335 | 80K | |
| Cloudflare AI Gateway | workers-ai/@cf/meta/llama-3.2-3b-instruct | global | $0.051 | $0.34 | 128K | |
| Pioneer | meta-llama/Llama-3.2-3B | global | $0.10 | $0.10 | 131.1K | |
| Pioneer | meta-llama/Llama-3.2-3B-Instruct | global | $0.10 | $0.335 | 131.1K | |
| Venice AI | llama-3.2-3b | global | $0.15 | $0.60 | 128K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Llama 3.2 3B Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
| Llama 3.2 3B Instruct | 0.0 | 3.2B | 128K | 128K | No | Llama 3.2 Community License |
What is Llama 3.2 3B?
Key information about Llama 3.2 3B and its available data.
2 3B Instruct is a large language model that supports a context length of 128K tokens and are state-of-the-art in their class for on-device use cases like summarization, instruction following, and rewriting tasks running locally at the edge.
Data as of 2026-08-11.
Llama 3.2 3B vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to Llama 3.2 3B
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
ME
Llama 3.2 11B
Meta
ME
Llama 3.1 8B
Meta
ME
Llama 3.2 90B
Meta
ME
Llama 3.1 70B
Meta
ME
Llama 4 Scout
Meta
ME
Llama 3.3 70B
Meta
FAQ
Common questions about Llama 3.2 3B.
When was Llama 3.2 3B released?
Llama 3.2 3B's default version was released on Sep 25, 2024.
How much does Llama 3.2 3B cost?
No official standard PAYG price is currently available for Llama 3.2 3B. The lowest tracked third-party offer starts at $0.02 input and $0.02 output via Inference.
Who created Llama 3.2 3B?
Llama 3.2 3B was created by Meta.
What is the context window for Llama 3.2 3B?
The default version has a 128K token context window.
Is Llama 3.2 3B open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer Llama 3.2 3B?
11 provider offerings are linked to the default version.
What models should I compare Llama 3.2 3B with?
Nearby ranked alternatives include DeepSeek-VL2, Gemma 3n E2B, Mistral NeMo.