Aleph Alpha / Luminous
Research Luminous Base variant pruned from 13B to 2.6B retained parameters for Graphcore IPU hardware.
Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.
Announcement dates and SDK selector suffixes do not establish exact checkpoint release dates.
Retained non-zero parameters after approximately 80% pruning; dense parent has 13B.
SDK caches tokenizers and model metadata locally; this does not establish prompt-cache support or discounted cached input.
No current official model-specific rate verified. Historical third-party dollar estimates are not treated as provider prices.
Reported relative demonstration measurements, not universal production guarantees.
Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.
Not collectedNo native selector verified.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Current retention and training-use commitments not established for this historical model record.
Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.
The system processes eligible input and retains reusable state.
Reuse depends on matching input, session state and the cache policy.
Reusing state can reduce repeated work. Any charges follow the billing rules below.
No provider-documented prompt cache verified for this exact model.
Local SDK tokenizer/context-size memoization is not provider prompt caching.
No cached-input rate verified.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 14 Oct 2026No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
Provider documents this Luminous model as a hosted commercial model or research demonstration. No original Luminous weight release was found in reviewed official model/repository listings. This is a qualified review finding; public SDK code and other open-weight Aleph Alpha families do not release these weights.
Classification evidence (opens in a new tab)