Parasail Pricing
Per-token pricing for 48 open-source LLMs on Parasail — a privacy-focused serverless inference network with on-demand deployments and batch processing.
Last updated: Nov 21, 2026
Get our weekly newsletter on pricing changes, new releases, and tools.
Parasail free tier & credits
No ongoing free tier — see credits & cheapest models — see the full Parasail free tier guide
Provider | Model | Context | Quant | Input/1M | Output/1M |
|---|---|---|---|---|---|
131k | fp8 | $0.030 | $0.030 | ||
131k | fp4 | $0.030 | $0.150 | ||
131k | bf16 | $0.040 | $0.050 | ||
S1 | 8k | bf16 | $0.040 | $0.050 | |
131k | bf16 | $0.050 | $0.330 | ||
131k | fp8 | $0.080 | $0.450 | ||
GR | 4k | fp16 | $0.080 | $0.110 | |
131k | bf16 | $0.090 | $0.300 | ||
128k | bf16 | $0.100 | $0.200 | ||
131k | fp4 | $0.100 | $0.750 | ||
QW | 262k | bf16 | $0.100 | $0.250 | |
QW | 262k | fp8 | $0.100 | $0.900 | |
QW | 262k | fp8 | $0.100 | $1.100 | |
QW | 262k | bf16 | $0.120 | $0.800 | |
262k | bf16 | $0.130 | $0.400 | ||
DS | 1049k | fp8 | $0.140 | $0.280 | |
DS | deepseek-ai/DeepSeek-V4-Flash-0731 | 1049k | fp8 | $0.140 | $0.280 |
QW | 131k | fp8 | $0.140 | $0.800 | |
262k | fp8 | $0.150 | $0.400 | ||
MM | 1049k | fp8 | $0.150 | $0.600 | |
131k | fp4 | $0.150 | $0.750 | ||
QW | 262k | fp8 | $0.150 | $1.000 | |
Z | 1049k | fp8 | $0.150 | $0.500 | |
Z | 1049k | fp4 | $0.188 | $0.625 | |
QW | 131k | fp8 | $0.210 | $1.900 | |
131k | fp8 | $0.220 | $0.500 | ||
QW | 262k | fp8 | $0.240 | $2.200 | |
QW | 262k | bf16 | $0.250 | $0.750 | |
DS | 1049k | fp8 | $0.300 | $1.200 | |
DS | deepseek/deepseek-v4.1-flash-1017 | 1049k | fp8 | $0.300 | $1.200 |
MM | 1049k | fp8 | $0.300 | $1.200 | |
131k | bf16 | $0.300 | $0.500 | ||
524k | fp8 | $0.350 | $1.000 | ||
1024k | bf16 | $0.400 | $0.400 | ||
DS | 1049k | fp8 | $0.450 | $1.800 | |
MM | 1049k | fp8 | $0.450 | $1.800 | |
QW | 262k | fp8 | $0.500 | $3.600 | |
33k | fp8 | $0.550 | $0.800 | ||
262k | int4 | $0.750 | $3.500 | ||
QW | 128k | fp8 | $0.800 | $1.000 | |
DS | deepseek/deepseek-v4-pro-0813 | 1049k | fp8 | $1.320 | $3.960 |
Z | 262k | fp4 | $1.400 | $4.400 | |
Z | 1049k | fp4 | $1.400 | $4.400 | |
DS | deepseek/deepseek-v4-pro-0813 | 1049k | fp8 | $1.980 | $5.940 |
1049k | fp4 | $2.000 | $10.000 | ||
Z | 1049k | fp4 | $2.100 | $6.600 | |
1049k | fp4 | $3.000 | $15.000 | ||
1049k | fp4 | $4.500 | $22.500 |
Pricing for Parasail endpoints aggregated via OpenRouter. Prices per 1M tokens.
About Parasail
Parasail is a serverless inference network offering pay-per-token access to open-source models, plus on-demand dedicated deployments and batch processing for large jobs. It is marketed on privacy — no data retention — and exposes an OpenAI-compatible API.
Many models are served at fp8, fp4 or int4 quantization to keep costs low. The table below shows the models Parasail currently serves with public per-token pricing, the quantization each price applies to, and context length.
Parasail Pricing FAQ
How much does Parasail cost?
Parasail charges per token, with input prices for smaller open models starting around $0.04–$0.10 per million tokens. See the live table above for current per-model input and output prices.
Does Parasail retain my data?
Parasail markets itself on privacy with a no-data-retention policy on its serverless inference. Check their current terms for the authoritative guarantee.
What is the difference between serverless and dedicated on Parasail?
Serverless is pay-per-token with no commitment. Dedicated deployments give you reserved capacity for steady or latency-sensitive workloads, and batch processing handles large offline jobs at lower cost.
Is Parasail OpenAI-compatible?
Yes. Parasail exposes an OpenAI-compatible API, so switching from another provider is usually just a base-URL and API-key change.
