Fine-Tuning Cost Calculator
Estimate the cost of fine-tuning a model on your data.
With the default inputs, $48.00 to fine-tune — 2,000,000 tokens x 3 epochs. Enter your own numbers below to recompute instantly; the full step-by-step math is shown under the worked example.
Results update automatically as you type.
- • Fine-tuned models also bill higher per-token inference rates than the base model.
- Training tokens
- 2,000,000
- Epochs
- 3
- Training price / 1M
- $8.00
- Total training cost
- $48.00
Estimate what a fine-tuning run will cost based on your dataset size, number of epochs and the provider's training price per million tokens.
How this is calculated
Training cost = training tokens × epochs × training price per million tokens. Fine-tuning bills by tokens processed across passes at a training rate that differs from — and is usually higher than — inference pricing.
Is this a good result? What to do next
A single small fine-tune is often only tens of dollars; cost climbs with dataset size and epochs. Just as important, fine-tuned models frequently carry higher inference rates, so weigh the one-time training cost against the ongoing serving cost.
Typical planning ranges
- Typical small run
- tens of dollars
- Cost scales with
- tokens × epochs
- Fine-tuned inference
- often pricier than base
Ranges are typical planning figures to sanity-check your result, not authoritative benchmarks. Your numbers will vary with use case, volume, and vendor.
How to improve this number
- Use fewer epochs and a curated, deduplicated dataset.
- Start small and evaluate before scaling the run.
- Prefer LoRA/adapter tuning where the provider offers it.
Common mistakes
- Ignoring the higher post-fine-tune inference pricing.
- Over-training with too many epochs.
When to use a different approach
To decide if it's worth it at all, use the fine-tune vs prompt calculator. For the ongoing inference bill, use the LLM API cost calculator.
Worked example (defaults)
With the default inputs above, here is the result:
- • Fine-tuned models also bill higher per-token inference rates than the base model.
- Training tokens
- 2,000,000
- Epochs
- 3
- Training price / 1M
- $8.00
- Total training cost
- $48.00
- cost = training tokens x epochs / 1e6 x price per 1M
Sources & references
Frequently asked questions
How is fine-tuning billed?+
Typically by training tokens processed × number of epochs, at a per-million-token training rate that differs from inference pricing.
Is inference more expensive after fine-tuning?+
Usually yes — fine-tuned models often carry higher per-token inference rates than the base model.
Related calculators
Answering a real question?
This calculator powers these problem-solving guides: