AutoInfra launches Tinker-compatible training API at lower cost
The pricing gap against Tinker suggests infra cost, not algorithms, is becoming a competitive lever for agentic research pipelines. Isolated algorithm, training and sandbox environments are built specifically to guard against reward hacking during self-improvement cycles, a design detail teams running autonomous R&D agents will study closely. Practitioners can benchmark the API against their current training stack before committing budget.
SF Research announced AutoInfra, an infrastructure platform for running rapid, parallel training experiments on shared GPU capacity. Experiments flow through a single Tinker-compatible training API, while an agent called TrainForge decides which experiments to run. AutoInfra also trains the models behind TrainForge and AutoR&D-Engineer, feeding agent discoveries into the next model version. GPT-OSS-120B LoRA training costs $0.199 per million tokens on AWS versus $2.33 at Tinker's listed price, a 91.5% reduction that saves an estimated $2.13M per 1T training tokens.
AutoInfra: Infrastructure for rapid, parallel experimentation
https://t.co/DxT05IqobY
TrainForge decides what to try. AutoInfra carries each experiment from request to result across shared GPU capacity, through a single Tinker-compatible training API.
It also trains the models that power TrainForge and AutoR&D-Engineer, so what those agents discover feeds directly into the next model version.
→ GPT-OSS-120B LoRA training at $0.199 per million tokens on AWS, compared with $2.33 at Tinker's listed price → 91.5% lower cost per token, an estimated $2.13M saved per 1T training tokens → Isolated algorithm, training, and sandbox environments guard against reward hacking across self-improvement cycles
#FutureOfAI #EnterpriseAI
