← All models
Balanced — strong quality at a mid priceOpen Weights
Switchyard Pricing
NVIDIA · nvidia-switchyard
Input / 1M tokens
$-1000000.000
Output / 1M tokens
$-1000000.000
Context window
1M tokens
≈ 1,333 pages of text
Max output
—
What it costs in practice
Typical request (1,200 in + 400 out tokens)$-1,600
1,000 requests / month$-1,600,000/mo10,000 requests / month$-16,000,000/mo100,000 requests / month$-160,000,000/moEstimate your own workloadOpens the calculator with Switchyard preloaded — adjust volume and token counts there.
Price history
Only one price point recorded so far.
The staircase chart appears once a price change is detected.
What it excels at
Large 1M-token context supports document-scale and long-sequence tasks. Open weights enable local deployment and customization without per-token fees.
The business tradeoff
Unknown benchmark performance and lack of managed API may increase integration and reliability overhead compared to established providers.
Head-to-head
Vendor list rates, as of Oct 2, 2026 · source: openrouter · per-request examples assume 1,200 input + 400 output tokens.