Model · reax-1.3
REAX 1.3
REAX is a decentralised lab on Bittensor subnet 92: miners train decision models in an open competition, and our releases build on the best of them. REAX 1.3 has two parts that run in parallel on one GPU; each part builds on a fine-tune trained by a REAX miner (uid 18 and uid 83). Jev-compatible interface, with a probability for every option.
| Model id | reax-1.3 |
|---|---|
| Tier | EU only (S1-Region: eu); Peer-to-Peer is not offered |
| Price | €0.400 or $0.400 per 1M input tokens, output free |
| Input billing | REAX 1.3 runs two models on every request, and input tokens count both. The first also reads inputs up to about 3,000 tokens twice, so a 2,600-token state bills about 8,100 tokens. usage.input_tokens shows the exact figure. |
| Questions per request | 1 |
| Input per question | 28,672 billed tokens including the state, not the length of your text: both models read the text, so about 78,000 characters of English prose or 17,000 of invoice-style lines fit (measured on sample text). Longer inputs are refused, not truncated |
| Question types | noul, choice, score; text only |
| Weights | decisionmodels/REAX-1.3 (Apache-2.0) |
| Inference code | REAX Engine (MIT), reax13 mode |
The two parts, their lineage, licences and notices for every contributor, including the miners, are in the model card. "Jev-compatible" describes the request/response interface only; it does not imply endorsement or identical outputs.
First request
curl
curl https://api.decisionmodels.io/v1/systemone \
-H "Authorization: Bearer $DECISIONMODELS_API_KEY" \
-H "Content-Type: application/json" \
-H "S1-Region: eu" \
-d '{"model":"reax-1.3","state":"A parcel arrived late.","questions":{"route":{"type":"choice","instructions":"Which team should handle the complaint?","criteria":{"delivery":"Delivery support","billing":"Billing support"}}}}'