Llama 3.1 8B — Spezifikationen
| Entwickler | Meta |
|---|---|
| Typ | LLM (dicht) |
| Modality | Text → Text |
| Parameter | 8B |
| Kontextfenster | 128 K |
| Maximale Ausgabe | — |
| Lizenz | Llama 3.1 Community (offen) |
| Offene Gewichte | Ja |
| Veröffentlicht | 2024 |
| Eingabepreis | 0,02 $ pro 1 Mio. |
| Ausgabepreis | 0,03 $/1 Mio. |
| API-Anbieter | Together, DeepInfra, OpenRouter, Ollama |
Lokal ausführen
| VRAM (4-Bit) | ~5 GB |
|---|---|
| Mindest-GPU | Jede 8-GB-GPU |
What is Llama 3.1 8B?
Llama 3.1 8B is the workhorse small Llama — 8B parameters, a 128K context, and about 5 GB
of VRAM at 4-bit, which runs on essentially any modern 8 GB GPU. It is one of the most widely
deployed open models in production and remains an excellent cheap baseline.
Its enduring value is less about capability than about ubiquity. Because it has been in
production for so long across so many stacks, it is the model with the deepest tooling
support, the most fine-tunes, the most quantised variants and the most known-good deployment
recipes — Ollama, vLLM, llama.cpp, Together, DeepInfra and OpenRouter all serve it without
surprises. That makes it the right choice for the first version of almost any local
deployment: get the pipeline working against a model that definitely runs, measure real
quality on your own data, then decide whether you need something larger. At $0.02 in / $0.03
out per million tokens the hosted price is close to a rounding error, so it also works well
as the cheap tier of a routing setup that escalates only when needed.
Llama 3.1 8B pricing: API cost per 1M tokens
| Eingabe (pro 1 Mio. Token) | $0.0200 |
|---|---|
| Ausgabe (pro 1 Mio. Token) | $0.0300 |
| Output/input ratio | 1.5× |
| Blended (4:1 in:out) | $0.0220 per 1M tokens |
What Llama 3.1 8B costs per month
Real monthly spend at a 4:1 input-to-output mix — the ratio a typical chat or RAG workload actually produces.
| Workload | Tokens/Monat | Cost / month |
|---|---|---|
| Nebenprojekt | 1 Mio. Eingabe / 0,25 Mio. Ausgabe | $0.03 |
| Kleines Team | 20 Mio. Eingabe / 5 Mio. Ausgabe | $0.55 |
| Produktion | 200 Mio. Eingabe / 50 Mio. Ausgabe | $5.50 |
Run your own numbers in the KI-API-Kostenrechner.
Self-host or pay the API?
Llama 3.1 8B is open-weight, so you can run it yourself. It needs ~5 GB of VRAM at 4-bit (Any 8GB GPU). Self-hosting only beats the API once your volume is high enough to keep that hardware busy — the Selbsthosting-vs.-API-Rechner works out the break-even point for your token volume.
Häufig gestellte Fragen
How much does Llama 3.1 8B cost per 1M tokens?
Llama 3.1 8B costs $0.0200 per 1M input tokens and $0.0300 per 1M output tokens. At a typical 4:1 input-to-output mix that blends to about $0.0220 per 1M tokens.
How much does Llama 3.1 8B cost per month?
A small-team workload of 20M input and 5M output tokens a month costs about $0.55 on Llama 3.1 8B. A side project (1M in / 0.25M out) costs roughly $0.03.
Can I run Llama 3.1 8B locally?
Yes. Llama 3.1 8B is open-weight and needs about ~5 GB of VRAM at 4-bit quantisation (Any 8GB GPU).
Why does Llama 3.1 8B charge more for output than input?
Output tokens are generated one at a time and cannot be batched the way a prompt can, so they cost the provider more to serve. Llama 3.1 8B charges 1.5× more for output, which is why prompt-heavy workloads are far cheaper to run than generation-heavy ones.
Prices are the published list rates for the model's primary API and are reviewed as providers change them. Volume, batch and cached-input discounts are not included. Compare every model side by side in the Datenbank für KI-Modelle oder das LLM-Leaderboard.
