Large
ambient/large
spread (output, p90 ÷ p10)
— too few providers
$1.50 – $2.00 · 2 providers · max÷min 1.3×
Cheapest / median / spread use providers that serve the model on their own capacity: first-party lab, hyperscaler, inference host.
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 |
|---|---|---|---|---|
Mistral first-party lab | $0.50 | $1.50 | $0.05 | $0.30 |
Ambient inference host | $0.60 | $2.00 | $0.15 | $0.44 |
Kilo Gateway router / aggregator | $2.00 | $6.00 | $0.20 | $1.22 |
OpenRouter router / aggregator | $2.00 | $6.00 | $0.20 | $1.22 |
NanoGPT router / aggregator | $2.01 | $6.00 | $0.20 | $1.22 |
AI/ML API unclassified | $2.12 | $6.35 | $0.21 | $1.29 |
Subscription plans (flat fee, not per-token) · 1
Flat-fee plan — the per-token price is bundled into a subscription, so $0/token is not free. Excluded from cheapest, median and spread.
- DevPass (LLM Gateway) mistral-large-latestsubscription plan
Free tiers (rate-limited) · 1
No payment required; rate limits and quotas apply (not tracked yet). Excluded from cheapest, median and spread.
- Kenari mistral-large:freefree tier
Data: models.dev (MIT), snapshot 2026-10-26. Prices are vendor-published; verify before purchase. Methodology.