Canonical model family identity
developers.openai.com
Snapshot source-snapshot-d1dba421db410a73bf01dc276e207774e841d5929f6bbb7b64f271f32b43cb52 · d1dba421db410a73bf01dc276e207774e841d5929f6bbb7b64f271f32b43cb52
Versioned model-family evidence
Reviewed public profile
GPT-OSS 20B currently has reviewed, calculated evidence for 1 of 4 capability and task scopes in the current release. Unsupported scopes remain explicitly marked rather than estimated.
Reviewed model profile
GPT-OSS 20B is included in Spring Prompt's current-model profile cohort. The linked provider page verifies the canonical family identity; it is not treated as benchmark evidence.
The current release records 1 calculated, 3 insufficient-evidence and 0 withheld results across 4 capability and task scopes. An insufficient-evidence result is not a zero score.
Scores on this profile come only from reviewed, source-exact configurations in the public catalogue. Missing coverage is not inferred from provider claims or neighbouring model families.
Cited release facts
Canonical model family identity
Snapshot source-snapshot-d1dba421db410a73bf01dc276e207774e841d5929f6bbb7b64f271f32b43cb52 · d1dba421db410a73bf01dc276e207774e841d5929f6bbb7b64f271f32b43cb52
Current immutable release
A precise 0–100 score is shown only where reviewed evidence supports one. Confidence describes evidence strength separately; an insufficient-evidence row is not a zero.
Capability
Score
Insufficient evidence
Capability
Score
73.173539/100
Score
Insufficient evidence
Task
Score
Insufficient evidence
Identity and provenance
gpt-oss-20b
Configuration configuration-sob-8beb522273f97432467c03e4f22dc0c07bc95abf7066ccafd61de09fa6e8e3fc · Representative
Reviewed 8 Aug 2026
Selected the exact reviewed product-family configuration from the pinned SOB release.
8beb522273f97432467c03e4f22dc0c07bc95abf7066ccafd61de09fa6e8e3fc
Traceable inputs
sob-v1
Metric: sob-overall