Winding down. Closed to new users; existing users can still create jobs until Jan 6, 2027. 4 offerings recorded: SFT, DPO, RFT and Vision SFT. Training is priced per 1M training tokens, per training hour and billed at another rate, in OpenAI's own units. Weight export: Not stated.
Closed to new users; existing users can still create jobs until Jan 6, 2027.
“OpenAI is winding down the fine-tuning platform. The platform is no longer accessible to new users, but existing users of the fine-tuning platform will be able to create training jobs for the coming months.”
New fine-tuning jobs end for active existing customers on Jan 6, 2027
“Active existing customers will no longer be able to create new fine-tuning jobs on this date. Inference on fine-tuned models will be disabled only when the underlying base model is deprecated.”
Winding down: closed to new users; existing users can still create jobs until Jan 6, 2027.
“OpenAI is winding down the fine-tuning platform. The platform is no longer accessible to new users, but existing users of the fine-tuning platform will be able to create training jobs for the coming months.”
Pricing table also lists gpt-4o-2024-08-06 and gpt-4o-mini-2024-07-18 (SFT guide's model list does not). gpt-4.1-nano fine-tunes are listed for removal on Oct 23, 2026 even though the SFT guide still lists the base. Hosting: no separate hosting fee is stated; tuned models are billed per token. Tuned-model inference prices are per 1M tokens (Standard tier); a cheaper 'data sharing' rate exists for o4-mini and Batch rates are lower.
Winding down: closed to new users; existing users can still create jobs until Jan 6, 2027.
“OpenAI is winding down the fine-tuning platform. The platform is no longer accessible to new users, but existing users of the fine-tuning platform will be able to create training jobs for the coming months.”
The pricing page has no separate DPO training row; it is not stated whether DPO uses the SFT per-token rate. DPO is text-only: 'DPO is currently only supported for text inputs and outputs.'
Tuned-model inference
Not stated
Hosting requirement
Not stated
Export weights
Not stated
Data limits
one-turn conversations only
“Currently, we only train on one-turn conversations for each example, where the preferred and non-preferred messages need to be the last assistant message.”
Winding down: closed to new users; existing users can still create jobs until Jan 6, 2027.
“OpenAI is winding down the fine-tuning platform. The platform is no longer accessible to new users, but existing users of the fine-tuning platform will be able to create training jobs for the coming months.”
The only RFT base, o4-mini-2025-04-16, and its fine-tunes (ft-o4-mini-2025-04-16) are listed for shutdown on October 23, 2026, five days after this snapshot.
Winding down: closed to new users; existing users can still create jobs until Jan 6, 2027.
“OpenAI is winding down the fine-tuning platform. The platform is no longer accessible to new users, but existing users of the fine-tuning platform will be able to create training jobs for the coming months.”
Training price taken from the gpt-4o-2024-08-06 fine-tuning row; the pricing page does not have a vision-specific row. gpt-4o-2024-08-06 is scheduled for shutdown Jan 14, 2027.
Research notes
Self-serve fine-tuning is being wound down: closed to organizations that never ran fine-tuning (May 7, 2026), closed to orgs with no fine-tuned-model inference in the past 60 days (July 2, 2026), and closed to all new jobs on Jan 6, 2027. Fine-tuned models keep serving until their base model is retired. Several tunable bases have announced shutdowns (see deprecations). Pricing table header says 'Prices per 1M tokens.' (training price is per 1M training tokens except o4-mini RFT, which is per hour).
Plain language
The methods, explained
SFTSupervised fine-tuning
You show the model example prompts with the answers you want, and it learns to answer like that. The default method almost everywhere.
You give pairs of answers to the same prompt, one preferred and one not, and the model learns to lean toward the preferred kind. Good for tone and style.
RFTReinforcement fine-tuning
The model attempts tasks and a grader (your code, or another model) scores each attempt; training pushes it toward higher-scoring answers. Often billed by training hour, plus the grader's tokens.
Vision SFTVision fine-tuning
Supervised fine-tuning where the examples include images, so the model gets better at a visual task.
DistillationDistillation
A large teacher model writes answers that are used to fine-tune a smaller, cheaper student model. You pay for the teacher's calls as well as the training.
LoRALoRA (adapter) fine-tuning
Trains a small set of extra weights, an adapter, on top of the frozen base model. Cheaper and faster than full fine-tuning; the adapter is served alongside the base model.
FullFull-parameter fine-tuning
Updates every weight in the model. Costs more than LoRA and produces a complete copy of the model.
CPTContinued pre-training
Keeps training the model on a large body of raw, unlabeled text from your domain, so it absorbs vocabulary and facts before any instruction tuning.
FAQ
Does OpenAI offer fine-tuning?
Winding down: closed to new users. Closed to new users; existing users can still create jobs until Jan 6, 2027. The source says: "OpenAI is winding down the fine-tuning platform. The platform is no longer accessible to new users, but existing users of the fine-tuning platform will be able to create training jobs for the coming months."
How much does OpenAI fine-tuning cost?
Per 1M training tokens: gpt-4.1-2025-04-14 $25.00, gpt-4.1-mini-2025-04-14 $5.00, gpt-4.1-nano-2025-04-14 $1.50, gpt-4o-2024-08-06 $25.00 and 2 more. Per training hour: o4-mini-2025-04-16 $100.00. Billed at another rate: grader tokens billed at the grader model's per-token rate. Not stated for Direct preference optimization. Prices are in OpenAI's own units and are not converted.
Can you download a model you fine-tune on OpenAI?
Not stated. The pages we checked do not say whether tuned weights can be exported.
Last verified . This is a directory of what each provider's own pages say, not a quality ranking. Prices are copied in the provider's own unit and never converted, so rows in different units are not comparable.
Newsletter
Know when a provider changes its fine-tuning terms
One email per week. New model releases, deprecations and pricing changes, and what they mean for what you build.