⚙ Curated ranking

Cheapest fine-tunable AI models

Fine-tuning lets you customize a model's behavior on your own data — often the highest-leverage way to improve task accuracy without prompt engineering. Ranked by input price ascending; fine-tuning training costs are separate and not factored here.

Methodology

How this ranking is built

Active models that support fine-tuning and have a published inference input price. Ranked from cheapest inference cost upward. Training costs are billed separately by each provider and aren't included here.

# Model Provider Input price (inference) Lifecycle Verified
1
gpt-oss-20b
OpenAI · Active
OP OpenAI $0.03/M input Active · 1h ago
2
gpt-oss-120b
OpenAI · Active
OP OpenAI $0.04/M input Active · 1h ago
3
GPT-4
OpenAI · Active
OP OpenAI $0.06/M input Active · 1h ago
4
Amazon Nova Lite
AWS Bedrock · Active
AB AWS Bedrock $0.06/M input Active · 1h ago
5
GPT-4o Mini
OpenAI · Active
OP OpenAI $0.15/M input Active · 1h ago
6
GPT-4.1 Mini
OpenAI · Active
OP OpenAI $0.4/M input Active · 1h ago
7
Amazon Nova Pro
AWS Bedrock · Active
AB AWS Bedrock $0.8/M input Active · 1h ago
8
o4-mini
OpenAI · Active
OP OpenAI $1.1/M input Active · 1h ago
9
GPT-4.1
OpenAI · Active
OP OpenAI $2/M input Active · 1h ago
10
GPT-4o
OpenAI · Active
OP OpenAI $2.5/M input Active · 1h ago
Refreshed as we re-verify each model against its provider's official docs. Every value links to the page it came from. Last updated Jun 15, 2026 07:59 UTC.