Frontier accuracy.
Small-model prices.
Build your first models with $200 free, then pay as you go. In production, pay per GPU hour for a dedicated endpoint, or run on your own infrastructure with a private deployment.
Pay as you go
Build your models
$200 free to start
Train a model and see the results on your own hardware.
- $200 on signup, no card needed
- Task description + as few as 10 examples to start
- Download the trained model and self-host it for R&D
- Community support on Slack
Production
Most popularHosted inference
~$0.04 / 1M tokens*
A dedicated GPU serving your model. You're billed per GPU hour of uptime, not per token; the per-token numbers are just what that works out to.
- Dedicated H100 endpoint running your distilled model
- Unlimited tokens, throughput is yours 24/7
- Effective cost around $0.04 / 1M tokens at sustained load
- Priority support
* Blended 3:1: three input tokens per output token.
Talk to us →Enterprise
Private deployments
Custom
Run distil labs models entirely inside your own cloud or on-premise.
- Deploy on your own infrastructure: your cloud, your VPC, or air-gapped
- Built for sensitive data: nothing leaves your environment
- Edge and on-device options down to 100M parameters
- Retraining from your own traces, on your own hardware
- SLAs, procurement and security review support
The same tokens, elsewhere
Effective price per 1M tokens at sustained load on one H100.
Blended price / 1M tokens*
Cheaper on every token. The 4B runs roughly 5× under Gemini Flash-Lite and 12× under GPT-5.4 Nano, and because you pay per GPU hour your effective rate falls further as traffic grows.
* Blended 3:1: three input tokens per output token. Effective rates at sustained load, derived from prefill and decode throughput on one H100.
Building a model
Pay as you go from a prepaid balance. Each job is priced at the tier of the model it runs, and larger models and thinking variants are in the higher tiers.
| Job | Low | Medium | High |
|---|---|---|---|
| Priced at the teacher's tier | |||
| For example | gpt-oss-120b | DeepSeek V4 Pro | Kimi K3 |
| Teacher evaluationper 1M test set tokens | $1.00 | $4.00 | $12.00 |
| Relabeling tracesper 1M trace tokens, for each teacher | $1.00 | $4.00 | $12.00 |
| Synthetic data generationper 1M generated tokens | $1.00 | $4.00 | $12.00 |
| Priced at the student's tier | |||
| For example | Qwen3 0.6B | Qwen3 4B | Qwen3.5 9B |
| SFT trainingper 1M train set tokens × epochs | $0.10 | $0.25 | $0.50 |
| RLVR trainingper 1M train set tokens × epochs | $0.60 | $1.50 | $3.00 |
| Deploymentper deployment | $5.00 | $5.00 | $5.00 |
Uploads, downloads and reads are free, and--dry-runshows a job's price before you start it.How billing works →
1
Train free
$200 to start, 10+ examples, a model in under 24 hours.
2
Deploy
Hosted endpoint, or your own hardware.
3
Improve
Production traces feed retraining automatically.
Questions
What happens at low traffic?+
A dedicated GPU makes sense above roughly 30% utilization. Below that, start on the free tier and self-host.
Which models can you distill into?+
Open-source students from 100M to ~9B parameters, including the Qwen3 family. We only use open-source teacher models, so your model is yours, with no licensing strings.
What does the $200 cover?+
Every new account starts with a $200 balance, no card needed. It pays for the jobs that build a model, at the prices above, and you can check a job's price before you start it. Uploads, downloads and local deployment are free.
Can we keep building without a production plan?+
Yes. Add funds and pay per job, with no subscription.Add funds →
Build your agent's first model free
Start from your agent's traces or a small set of labelled examples, with $200 on us. Then pay per job as you go.