Llama3 3 70B

meta-llama/llama3-3-70b

Meta· 131K ctx

open weightstool call quant: unknown

spread (output, p90 ÷ p10)

— too few providers

$2.80 – $2.80 · 1 providers · max÷min 1.0×

1 provider priced · 1 offering

Including routers & resellers

Cheapest / median / spread use providers that serve the model on their own capacity: first-party lab, hyperscaler, inference host.

Provider classes in the stats:

Blended $/M assumptionsInput : output: 1Cache-hit share of input%

blended = 90.9% input × [(1 − 70% cache) × input + 70% cache × cache-read] + 9.1% × output + cache writes (30% of uncached input) where a write price is published

A provider that publishes no cache price is blended with cache share 0 (all input at the input price) and labelled “no cache price published” — never penalised against identical in/out prices.

Provider	Input	Output	Cache rd	Blended
NanoGPT router / aggregator	$1.75	$2.75	$1.75	$1.84
Venice AI inference host	$0.70	$2.80	—	$0.89no cache price published

Connect

Provider (default: cheapest output among direct providers)

$2.80 /M · inference host

opencode.json

{
  "provider": {
    "venice": {
      "npm": "venice-ai-sdk-provider",
      "models": {
        "llama-3.3-70b": {}
      }
    }
  }
}

Set VENICE_API_KEY in your environment. Generated — verify against the OpenCode docs.

Output $/M over time7-day preview

Median line, min–max band. Pre-2026-09 points are catalog-recorded dates from models.dev git history, not provider effective dates.

Data: models.dev (MIT), snapshot 2026-09-20. Prices are vendor-published; verify before purchase. Methodology.