Llama 3.2 3B Instruct Meta · Released Sep 25, 2024 · Open weights
Small instruction-tuned Llama model for multilingual dialogue and agentic retrieval
Strongest at Writing Better than 0% of the past year's models · one benchmark so far
Price per 1M tokens $0.05 / $0.33 input / output · OpenRouter
Context window 131.1K tokens in · 131.1K out
Since the last one 64 days after Llama-3.1-8B-Instruct Capability profile Where it lands among 90 models released in the past year. 50 is the average model.
Writing
13 Better than 0% · 1 test
Reading & documents
No results yet
Visual reasoning
No results yet
Computer use
No results yet
Where these scores come from Each capability combines the published benchmarks listed here, rescaled so 50 is the reference average and 15 points is one standard deviation. A hollow marker rests on a single benchmark.
Price against capability The shaded corner holds anything better and cheaper. Prices blend input and output 3 to 1.
4 models from the past year score higher on Writing for less: DeepSeek V4 Flash , DeepSeek V4 Flash 0731 , Gemma 4 26B A4B IT and 1 more.
$0.1 $0.3 $1 $3 $10 20 40 60 Kimi K3.1 Pro: 70, $4 per 1M Claude Haiku 5.5: 46, $0.2 per 1M Mistral Large 4: 41, $2.06 per 1M Gemini 4 Argon: 63, $4 per 1M GPT-6.1 Sol: 64, $4 per 1M Claude Sonnet 5.5: 55, $4 per 1M GPT-6 Sol: 61, $4 per 1M GPT-6 Luna: 22, $0.2 per 1M Claude Opus 5.5: 65, $8 per 1M Grok 4.7: 64, $3 per 1M DeepSeek V4.1 Flash: 56, $0.27 per 1M GPT-6 Astra: 69, $20 per 1M Muse Spark 1.3: 65, $2 per 1M Gemini 3.8 Flash: 65, $1.5 per 1M Claude Fable 5.1: 69, $20 per 1M GLM-5.3-Flash: 15, $0.24 per 1M DeepSeek V4 Flash Vision Exp: 56, $0.32 per 1M Qwen3.8 27B: 52, $1.12 per 1M GLM-5.3: 65, $2.15 per 1M Gemini 3.7 Flash: 61, $1.5 per 1M DeepSeek V4 Pro 0813: 49, $0.99 per 1M Qwen3.8 2.4T A95B: 62, $3 per 1M Grok 4.6: 44, $3 per 1M Muse Glimmer 30B: 59, $0.52 per 1M Muse Spark 1.2: 62, $2 per 1M Qwen3.8 Max: 57, $3 per 1M DeepSeek V4 Flash 0731: 52, $0.067 per 1M Claude Opus 5: 64, $10 per 1M Gemini 3.6 Flash: 63, $1.5 per 1M Gemini 3.5 Flash Lite: 53, $0.85 per 1M Kimi K3: 64, $6 per 1M GPT-5.6 Terra: 57, $4.5 per 1M GPT-5.6 Sol: 65, $8 per 1M GPT-5.6 Luna: 53, $0.45 per 1M Grok 4.5: 44, $3 per 1M Claude Sonnet 5: 57, $4 per 1M GLM-5.2: 55, $2.15 per 1M Kimi K2.7 Code: 23, $1.71 per 1M Claude Fable 5: 68, $20 per 1M Claude Opus 4.8: 62, $10 per 1M Qwen3.7 Max: 48, $3.75 per 1M Gemini 3.5 Flash: 56, $3.38 per 1M DeepSeek V4 Pro: 54, $0.26 per 1M DeepSeek V4 Flash: 52, $0.11 per 1M GPT-5.5: 64, $11.2 per 1M Qwen3.6 27B: 16, $1.35 per 1M Kimi K2.6: 56, $1.71 per 1M Qwen3.6 Max Preview: 44, $2.92 per 1M Grok 4.3: 29, $1.56 per 1M Grok Build 0.1: 43, $1.25 per 1M Claude Opus 4.7: 61, $10 per 1M Muse Spark 1.1: 58, $2 per 1M GLM-5.1: 56, $2.15 per 1M Gemma 4 31B IT: 54, $0.25 per 1M Gemma 4 26B A4B IT: 51, $0.11 per 1M Qwen3.6 Plus: 31, $1.12 per 1M GPT-5.4 nano: 49, $0.46 per 1M GPT-5.4 mini: 52, $1.69 per 1M GPT-5.4: 63, $5.62 per 1M Gemini 3.1 Flash Lite Preview: 43, $0.56 per 1M Qwen3.5 35B-A3B: 39, $0.69 per 1M Qwen3.5 27B: 52, $0.82 per 1M Qwen3.5 122B-A10B: 47, $1.1 per 1M Gemini 3.1 Pro Preview: 57, $4.5 per 1M Claude Sonnet 4.6: 57, $6 per 1M Qwen3.5 397B-A17B: 59, $1.35 per 1M GLM-5: 60, $1.55 per 1M Claude Opus 4.6: 56, $10 per 1M Kimi K2.5: 62, $1 per 1M GLM-4.7: 58, $1 per 1M Gemini 3 Flash Preview: 49, $1.12 per 1M GPT-5.2 Codex: 46, $4.81 per 1M GPT-5.2: 59, $4.81 per 1M Mistral Large 3: 38, $0.75 per 1M Ministral 3 14B: 28, $0.2 per 1M DeepSeek V3.2: 49, $0.32 per 1M Kimi K2 Thinking: 61, $1.07 per 1M Llama-3.1-8B-Instruct: 17, $0.058 per 1M Llama 3.2 3B Instruct: 13, $0.12 per 1M Llama 3.2 3B Instruct Price per 1M tokens → ↑ Writing $0.1 $1 $10 20 40 60 Kimi K3.1 Pro: 70, $4 per 1M Claude Haiku 5.5: 46, $0.2 per 1M Mistral Large 4: 41, $2.06 per 1M Gemini 4 Argon: 63, $4 per 1M GPT-6.1 Sol: 64, $4 per 1M Claude Sonnet 5.5: 55, $4 per 1M GPT-6 Sol: 61, $4 per 1M GPT-6 Luna: 22, $0.2 per 1M Claude Opus 5.5: 65, $8 per 1M Grok 4.7: 64, $3 per 1M DeepSeek V4.1 Flash: 56, $0.27 per 1M GPT-6 Astra: 69, $20 per 1M Muse Spark 1.3: 65, $2 per 1M Gemini 3.8 Flash: 65, $1.5 per 1M Claude Fable 5.1: 69, $20 per 1M GLM-5.3-Flash: 15, $0.24 per 1M DeepSeek V4 Flash Vision Exp: 56, $0.32 per 1M Qwen3.8 27B: 52, $1.12 per 1M GLM-5.3: 65, $2.15 per 1M Gemini 3.7 Flash: 61, $1.5 per 1M DeepSeek V4 Pro 0813: 49, $0.99 per 1M Qwen3.8 2.4T A95B: 62, $3 per 1M Grok 4.6: 44, $3 per 1M Muse Glimmer 30B: 59, $0.52 per 1M Muse Spark 1.2: 62, $2 per 1M Qwen3.8 Max: 57, $3 per 1M DeepSeek V4 Flash 0731: 52, $0.067 per 1M Claude Opus 5: 64, $10 per 1M Gemini 3.6 Flash: 63, $1.5 per 1M Gemini 3.5 Flash Lite: 53, $0.85 per 1M Kimi K3: 64, $6 per 1M GPT-5.6 Terra: 57, $4.5 per 1M GPT-5.6 Sol: 65, $8 per 1M GPT-5.6 Luna: 53, $0.45 per 1M Grok 4.5: 44, $3 per 1M Claude Sonnet 5: 57, $4 per 1M GLM-5.2: 55, $2.15 per 1M Kimi K2.7 Code: 23, $1.71 per 1M Claude Fable 5: 68, $20 per 1M Claude Opus 4.8: 62, $10 per 1M Qwen3.7 Max: 48, $3.75 per 1M Gemini 3.5 Flash: 56, $3.38 per 1M DeepSeek V4 Pro: 54, $0.26 per 1M DeepSeek V4 Flash: 52, $0.11 per 1M GPT-5.5: 64, $11.2 per 1M Qwen3.6 27B: 16, $1.35 per 1M Kimi K2.6: 56, $1.71 per 1M Qwen3.6 Max Preview: 44, $2.92 per 1M Grok 4.3: 29, $1.56 per 1M Grok Build 0.1: 43, $1.25 per 1M Claude Opus 4.7: 61, $10 per 1M Muse Spark 1.1: 58, $2 per 1M GLM-5.1: 56, $2.15 per 1M Gemma 4 31B IT: 54, $0.25 per 1M Gemma 4 26B A4B IT: 51, $0.11 per 1M Qwen3.6 Plus: 31, $1.12 per 1M GPT-5.4 nano: 49, $0.46 per 1M GPT-5.4 mini: 52, $1.69 per 1M GPT-5.4: 63, $5.62 per 1M Gemini 3.1 Flash Lite Preview: 43, $0.56 per 1M Qwen3.5 35B-A3B: 39, $0.69 per 1M Qwen3.5 27B: 52, $0.82 per 1M Qwen3.5 122B-A10B: 47, $1.1 per 1M Gemini 3.1 Pro Preview: 57, $4.5 per 1M Claude Sonnet 4.6: 57, $6 per 1M Qwen3.5 397B-A17B: 59, $1.35 per 1M GLM-5: 60, $1.55 per 1M Claude Opus 4.6: 56, $10 per 1M Kimi K2.5: 62, $1 per 1M GLM-4.7: 58, $1 per 1M Gemini 3 Flash Preview: 49, $1.12 per 1M GPT-5.2 Codex: 46, $4.81 per 1M GPT-5.2: 59, $4.81 per 1M Mistral Large 3: 38, $0.75 per 1M Ministral 3 14B: 28, $0.2 per 1M DeepSeek V3.2: 49, $0.32 per 1M Kimi K2 Thinking: 61, $1.07 per 1M Llama-3.1-8B-Instruct: 17, $0.058 per 1M Llama 3.2 3B Instruct: 13, $0.12 per 1M Llama 3.2 3B Instruct Price per 1M tokens → ↑ Writing OpenRouter $0.05 / $0.33
Since Llama-3.1-8B-Instruct 5 releases in the Llama-3.3-70B-Instruct line.
Released 64 days later
Price in / out $0.05 / $0.33 +109%
Context 131.1K 1.0×
Writing 13 −4.1
A newer release is out: Llama-3.3-70B-Instruct .
Meta's past year 8 release days in the past year, about one every 46 days. The next is expected around Dec 13, 2026 . Forecasts ↗
Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Specifications RELEASE DATE Sep 25, 2024
WEIGHTS Open weights
CONTEXT WINDOW 131.1K
MAX OUTPUT 131.1K
REASONING Not listed
TOOL CALLING Not listed
INPUTS text
OUTPUTS text
Models.dev record ↗
Specifications and prices from Models.dev; availability and limits may differ by provider. Benchmarks from Epoch AI (CC BY 4.0), LiveBench and the other sources named above.
