Technical profile

A runtime built around the model.

Kernelwright applies a repeatable optimization discipline across model launches: profile the model and hardware together, then tune kernels, memory layout, quantization, KV cache, batching, and serving policy as one execution system. Performance is reported from exact-hardware, end-to-end measurement.

Model focusOpen weightNarrow catalog, deep optimization
Optimization unitFull stackKernel to admission control
IntegrationOpenAI APICompatible request surface

Interface

Compatibility contract

The serving interface is designed for marketplace routing and direct product integration, with model-specific capabilities declared explicitly.

CapabilityCommitmentStandard
Chat completionsOpenAI-compatible request and response shapesCore contract
StreamingServer-sent events with measured time to first tokenCore contract
Tool callingCompatible tool definitions and structured callsModel-qualified
Structured outputSchema-constrained response supportModel-qualified
UsagePrompt, completion, and total-token accountingCore contract
Admission controlEarly 429 responses instead of unbounded queuesCore contract

Measurement standard

What Kernelwright measures

Aggregate prompt and completion throughput across all requests

Time to first token and inter-token latency by concurrency level

Correctness across text, streaming, tools, and structured output

Queue delay, rate limits, errors, retries, and recovery behavior

Sustained thermal and memory behavior on the exact host

Serving cost and contribution at the observed request mix

Capacity policy

No benchmark theater.

Published capacity will reflect sustained validated throughput with normal compatibility features enabled. If demand exceeds that capacity, the endpoint will reject early rather than hide latency in a long queue.