Llama 3.1 8B API pricing

15 providers serve Llama 3.1 8B. Llama 3.1 8B is an open-weight model (Llama 3.1 Community) with 8B total parameters, up to 131K context. Prices below are per 1M tokens, cheapest input first (self-hosting math further down).

Refreshed 2026-08-23 · list prices from provider APIs
Input floor $0.020/1M at Nebius and Novita AI Output floor $0.030/1M at Nscale Find your workload winner →
Provider Input $/1M Output $/1M Cache read $/1M Context
Nebius $0.020 $0.060 - 128K Use
Novita AI $0.020 $0.050 - 16K Use
DeepInfra $0.030 $0.050 - 131K Use
Nscale $0.030 $0.030output floor - - Use
Llamagate $0.030 $0.050 - 131K
Vercel AI Gateway $0.050 $0.080 - 131K Use
OpenRouter marketplace quote $0.050 $0.080 $0.025 131K Use
FriendliAI $0.100 $0.100 - 8K Use
OVHcloud $0.100 $0.100 - 131K Use
SambaNova $0.100 $0.200 - 16K Use
Hyperbolic $0.120 $0.300 - 33K Use
Cloudflare $0.152 $0.287 - 32K Use
Perplexity $0.200 $0.200 - 131K Use
Azure AI Foundry $0.300 $0.610 - 128K Use
Oracle OCI $0.720 $0.720 - 128K Use

The cheapest way to run Llama 3.1 8B

On an illustrative 70M-input / 30M-output monthly workload, today's lowest listed cost is roughly $2.90/month, at Novita AI. Input-heavy, output-heavy and cache-heavy workloads can produce different winners. Use the live mix calculator below rather than combining floors from two different providers. On output-token cost alone, a busy self-hosted deployment can work out cheaper (~$0.014/1M output on V100). See the math below. Run your own numbers in the breakeven calculator, or read the breakeven math.

Your actual monthly cost

Uses today's list prices

Cache read uses the listed cache rate where available; otherwise it falls back to normal input price. Batch, write-cache, volume and negotiated discounts are excluded.

Or run it yourself: Llama 3.1 8B self-hosting economics

Llama 3.1 8B is open-weight (Llama 3.1 Community), so the API price above competes with the GPU-hour market. 8B total parameters needs roughly 11 GB VRAM at FP8 or 6 GB at INT4, KV-cache headroom included.

GPU that fits (FP8, single card)VRAMFrom $/hrCheapest at
V100 32 GB $0.022marketplace Vast.ai Rent
RTX 3090 24 GB $0.068marketplace Vast.ai Rent
RTX 5080 16 GB $0.102marketplace Vast.ai Rent
RTX 4090 24 GB $0.134marketplace Vast.ai Rent
RTX A5000 24 GB $0.160community RunPod Rent
RTX A4000 16 GB $0.170community RunPod Rent
Breakeven estimate: a well-batched vLLM deployment on one V100 ($0.022/hr) at ~844 aggregate tok/s and 50% utilization works out to roughly $0.014 per 1M output tokens, versus $0.030 via the cheapest output-token API (Nscale). Self-hosting wins on cost if you can keep the GPU busy. Planning estimate: throughput varies with hardware, quantization, batch size and context. Tune it in the calculator.

Related models

ModelCheapest in $/1MCheapest out $/1MProviders
Llama 3.3 70B $0.100 $0.200 15
Llama 4 Scout $0.050 $0.100 11
Llama 3.1 70B Instruct $0.120 $0.300 10
Llama 4 Maverick $0.050 $0.100 9
Llama 3.2 3B $0.020 $0.020 9
Llama-3-70b $0.120 $0.300 7

FAQ

What is the cheapest API for Llama 3.1 8B?

As of 2026-08-23, the lowest input price for Llama 3.1 8B, $0.020 per 1M tokens, is the same list price at Nebius and Novita AI. The lowest output price is Nscale at $0.030 per 1M output tokens. The cheapest provider for a real workload depends on its input/output mix. The next-lowest input price is DeepInfra at $0.030, a 50% difference.

How much VRAM do you need to self-host Llama 3.1 8B?

Llama 3.1 8B has 8B parameters, so plan for roughly 11 GB of VRAM at FP8 or 6 GB at INT4/AWQ, KV-cache headroom included. A single V100 (32 GB) fits it.

← All models · GPU rental prices · Breakeven calculator · Get this data as JSON