Pricing that follows your usage
Training
Pricing per
1M tokens
Training, sampling, and prefill.
Rates vary by model and context length.
Storage
Storage rate per
$0.10 / GB / month
Checkpoint storage for all models
Serverless inference Beta
Pricing per
1M tokens
Input, output, and cached input.
Available for Inkling and Inkling-Small.
Training & sampling
- Cached prefillWhen sampling, input tokens reused from the prompt cache cost 80% less.
- Forward passesForward-only requests from a training client are charged at the training rate.
- Mixture of expertsMixture-of-experts models are priced by active parameters, rather than total model size.
USD per 1M tokens
| Model | Context | Prefill | Cached prefill | Sample | Train |
|---|---|---|---|---|---|
 Inkling50% off64K | 64K | $1.87 | $0.374 | $4.68 | $5.61 |
Inkling256K | 256K | $1.87 | $0.374 | $4.68 | $11.23 |
Inkling-Small50% off64K | 64K | $0.58 | $0.116 | $1.44 | $1.73 |
Inkling-Small256K | 256K | $0.58 | $0.116 | $1.44 | $3.47 |
 Nemotron-3.5-Super-VL-120B-A12BNew50% off64K | 64K | $0.57 | $0.114 | $1.44 | $1.276 |
Nemotron-3.5-Super-VL-120B-A12B256K | 256K | $0.57 | $0.114 | $1.44 | $2.32 |
Nemotron-3.5-Lightning-30B-A3B50% off64K | 64K | $0.184 | $0.0368 | $0.495 | $0.44 |
Nemotron-3.5-Lightning-30B-A3B256K | 256K | $0.184 | $0.0368 | $0.495 | $0.80 |
Nemotron-3-Ultra-550B-A55B50% off64K | 64K | $1.80 | $0.36 | $6.225 | $5.478 |
Nemotron-3-Ultra-550B-A55B256K | 256K | $1.80 | $0.36 | $6.225 | $9.96 |
Nemotron-3-Super-120B-A12B50% off64K | 64K | $0.57 | $0.114 | $1.44 | $1.276 |
Nemotron-3-Super-120B-A12B256K | 256K | $0.57 | $0.114 | $1.44 | $2.32 |
Nemotron-3-Nano-30B-A3B64K | 64K | $0.195 | $0.039 | $0.495 | $0.44 |
 GLM-5.3256K | 256K | $3.78 | $0.756 | $12.15 | $14.58 |
GLM-5.3-FlashNew256K | 256K | $0.84 | $0.168 | $2.11 | $5.58 |
 Kimi-K2.632K | 32K | $2.205 | $0.441 | $5.49 | $4.84 |
Kimi-K2.6128K | 128K | $2.205 | $0.441 | $5.49 | $15.40 |
 Qwen3.8-27B64K | 64K | $0.6975 | $0.1395 | $5.595 | $4.103 |
Qwen3.8-27B256K | 256K | $0.6975 | $0.1395 | $5.595 | $7.46 |
Qwen3.6-35B-A3B64K | 64K | $0.45 | $0.09 | $1.335 | $1.177 |
Qwen3.6-35B-A3B256K | 256K | $0.45 | $0.09 | $1.335 | $2.14 |
Qwen3.6-27B64K | 64K | $1.86 | $0.372 | $5.595 | $4.103 |
Qwen3.5-397B-A17B64K | 64K | $1.65 | $0.33 | $7.50 | $6.60 |
Qwen3.5-397B-A17B256K | 256K | $1.65 | $0.33 | $7.50 | $12.00 |
Qwen3.5-35B-A3B-Base64K | 64K | $0.54 | $0.108 | $1.335 | $1.177 |
Qwen3.5-9B64K | 64K | $0.30 | $0.06 | $0.807 | $1.463 |
Qwen3.5-9B-Base64K | 64K | $0.30 | $0.06 | $0.807 | $1.463 |
Qwen3.5-4B64K | 64K | $0.33 | $0.066 | $1.005 | $0.737 |
Qwen3.5-4B256K | 256K | $0.33 | $0.066 | $1.005 | $1.34 |
Qwen3-8B32K | 32K | $0.195 | $0.039 | $0.60 | $0.44 |
 GPT-OSS-120B32K | 32K | $0.33 | $0.066 | $0.84 | $0.737 |
GPT-OSS-120B128K | 128K | $0.33 | $0.066 | $0.84 | $2.33 |
GPT-OSS-20B32K | 32K | $0.18 | $0.036 | $0.45 | $0.396 |
 DeepSeek-V4.1-FlashNew256K | 256K | $0.60 | $0.12 | $3.60 | $4.98 |
DeepSeek-V3.132K | 32K | $1.695 | $0.339 | $4.215 | $3.718 |
Serverless inferenceBeta
- BetaProduction offering coming soon. Not yet recommended for intensive production use.
- ModelsCurrently available for Inkling and Inkling-Small only.
USD per 1M tokens
| Model | Context | Input | Cached input | Output |
|---|---|---|---|---|
Inkling256K | 256K | $1.00 | $0.17 | $4.05 |
Inkling-Small256K | 256K | $0.30 | $0.06 | $1.20 |
Discount reflected in pricing.
Tinker artwork inspired by the TinkerToy Computer, invented by Danny Hillis and Brian Silverman