Meta (Llama)
Llama 3.1 8B
openrouter/llama-3.1-8bContext
131K
Input / 1M tokens
$0.0200
Output / 1M tokens
$0.0400
Cached input / 1M tokens
— (bills at input price)
Cache write / 1M tokens
— (bills at input price)
Prices checked 2026-10-13 19:27 UTC
About this model
Llama 3.1 8B
Meta's Llama 3.1 8B — the open-weights workhorse. Small, fast, permissively licensed, and capable enough for routing, classification, and short-form generation tasks.
Key facts
- Maker
- Meta (Llama)
- Context window
- 131K tokens
- Max output
- 118K tokens
- Lane
- Guarded: the provider has confirmed it does not train on your data
- Modality
- Text
- Input price
- $0.0200 per 1M tokens
- Output price
- $0.0400 per 1M tokens
- Processing region(s)
- us
- Route
- Via OpenRouter
- Quantization
- Served at fp8
- Cheapest plan
- Available from the Budget Pass plan
Released
2024-07
Training cutoff
2023-12
Parameters
8.0B (dense)
Ways to call
Ways to call Llama 3.1 8B
Each model ID below is callable through the Anoman AI API. Price and processing region depend on the route.
| Model ID | Route | Input / output per 1M tokens | Region(s) | Lane | Context |
|---|---|---|---|---|---|
| openrouter/llama-3.1-8b | Via OpenRouter | $0.0200 / $0.0400 | us | Guarded | 131K |
| llama-3.1-8b-openrouter-backup | Via OpenRouter | $0.0200 / $0.0400 | us | Guarded | 131K |
Best for
Use cases
- Request routing + classification in front of larger models
- High-throughput summarization / extraction
- Synthetic data generation
- On-device fallback for offline UX
Strengths
What it does well
- Permissive license — bundle in commercial products
- Multilingual (8 languages officially supported)
- Strong instruction-following for its size
- Massive community fine-tunes available if a specialist is needed
Limitations
Know the trade-offs
- Knowledge cutoff 2023 — relatively dated for fresh-fact queries
- Reasoning limited — not for multi-step or math-heavy workflows
- Tool calling weaker than Qwen 2.5 7B at similar size
Benchmarks
Published scores
Scores from official model cards and public leaderboards. Higher is better unless noted.
| Benchmark | Score | Measures |
|---|---|---|
| MMLU | 73.0 | General knowledge across 57 subjects |
| HumanEval | 68.6 | Python code generation, pass@1 |
| MATH | 49.0 | Mathematics, mixed difficulty |
Code example
Chat completion with Llama 3.1 8B
from openai import OpenAI
client = OpenAI(
base_url="https://api.anoman.io/v1",
api_key="anm-sk-..."
)
response = client.chat.completions.create(
model="openrouter/llama-3.1-8b",
messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)Alternatives
Similar models in our catalog
Same maker first, then the closest price to Llama 3.1 8B in the Anoman catalog.