AllocateApache 2.0 licensed language models, ranked by price
The catalog lists 17 Apache 2.0 licensed language models. The cheapest is OpenAI GPT-OSS 20B at $0.05 per million input tokens; the strongest large bases include Qwen 3.5 (397B MoE) and Qwen3 235B. Apache 2.0 permits commercial use, modification, and fine-tuning with no strings on the output.
Provider list prices from the Allocate catalog, checked 2026-07-08. Billed price on Allocate is list plus the 7% transaction fee.
Why the license leads the decision
Apache 2.0 is the license legal teams approve fastest: commercial use, modification, redistribution, and a patent grant, with no usage thresholds or field-of-use conditions. If you plan to fine-tune and own the result, starting from this list removes a whole category of review.
Ranking is by combined input and output list price. Capability does not rank neatly, so pick the size class your task needs and test: small models for classification and extraction, the large MoE bases for reasoning and generation.
Common questions
Can I fine-tune Apache 2.0 models commercially?
Yes. Apache 2.0 permits commercial use, modification, and distribution, with an explicit patent grant. Fine-tuned weights are yours; on Allocate they stay inside your isolation boundary.
What is the cheapest Apache 2.0 model?
OpenAI GPT-OSS 20B at $0.05 per million input tokens at list. On Allocate you pay list plus the 7% transaction fee.
Related
Every model here sits behind one key on Allocate: route by name, meter per route, and swap the model in one click.