ThermoIR
Transforms mission workloads, model graphs and numerical kernels into lower-energy execution plans that can target software, PB64, CLAS16 or hybrid backends.
OpenEnergyIR Compute
The compute stack behind autonomous spacecraft.
The same compiler, runtime, benchmark and proof layer that flies onboard is available on the ground today — for engineers and scientists running energy-bound workloads in the data centre.
Available now
Run it on the ground before it flies.
Spacecraft compute is the hardest case of a general problem: fixed power, fixed thermal headroom, and results that have to be trusted. The platform that solves it is open to developers today, with metered API access, published benchmarks and signed certificates for every run.
- Metered APIOpenAI-compatible endpoints, per-token and per-run billing, usage dashboards.
- Published benchmarksMeasured, simulated and projected results kept separate and independently checkable.
- Signed resultsEvery run returns a certificate recording what executed, where and at what fidelity.
Product suite
Compiler, runtime, benchmark and proof.
Six components, one execution path. Each links through to its full documentation on the developer platform.
ThermOS
Schedules work against power, heat, latency, fidelity and mission priority so compute can adapt to operational constraints.
OpenThermoBench
Keeps measured results, simulations and projections separated so performance claims remain inspectable and procurement-ready.
OpenThermoStack
Wraps compile, run, telemetry and reporting into one product surface for teams testing energy-aware computation on representative workloads.
OpenThermoBridge
Connects existing APIs, model pipelines and mission-software interfaces to EnergyIR execution without forcing a full rewrite.
OpenEcert / Veritable
Produces signed provenance for what ran, where it ran, which fidelity was selected and why the result can be checked later.
OpenDeveloper interface
Expose the stack like software. Preserve the evidence like infrastructure.
Developers should not need to understand every thermodynamic or hardware detail before they can test a workload. The interface stays familiar while the platform returns the energy, fidelity and proof records that aerospace teams need.
energyir.compute("cloud-triage")
.quality("mission-safe")
.budget(power="tight", latency="realtime")
.targets(["software", "PB64", "CLAS16"])
.run()
.return(["result", "energy", "proof"])Execution path
From workload to signed result.
Describe
Define the workload, target quality, energy budget, latency window and hardware envelope.
Compile
ThermoIR generates candidate execution plans across available software and EnergyIR backend paths.
Run
ThermOS selects the route that fits the current power, heat, timing and fidelity constraints.
Measure
ThermoBench reports timing, energy, data movement, quality and caveats alongside the result.
Certify
The proof layer records the evidence needed for mission review, audit or independent verification.
Space mapping
How the developer stack becomes mission infrastructure.
Ground development
Use familiar software interfaces to characterize workloads before they enter a flight integration path.
Mission integration
Freeze target execution routes, telemetry hooks, fallbacks and certificate requirements for spacecraft software.
Onboard operation
Run local inference, optimization and triage under constrained power and communications conditions.
Post-pass assurance
Review signed evidence, benchmark notes and caveats after the spacecraft has made autonomous decisions.
Mission fit
Request the EnergyIR Compute brief.
Use this path for teams evaluating compiler interfaces, developer access, benchmark context, API integration or proof records behind spacecraft workloads.
Trust signals
Built for mission environmentsDesigned for radiation-aware computeBenchmark-backed architectureExport-control aware collaboration