← Back to Explorer
Pending

GPT-OSS-20b (2025)

GPT-OSS-20b is a large language model released by OpenAI, released under the Apache-2.0 license, with a claimed knowledge cutoff of 2025-08-05. Its weights, configuration, and tokenizer files have been cryptographically hashed and are queued for on-chain attestation by Weight Registry.

Description

OpenAI's first-ever open-weight model release, under Apache 2.0 -- a genuine historical first for a lab that had otherwise kept every model closed/API-only since GPT-2.

Metadata

Type
LLM
Issuer
OpenAI
License
Apache-2.0
Size
12.82 GB
Knowledge cutoff
2025-08-05
Source
https://huggingface.co/openai/gpt-oss-20b
Revision
6cee5e81ee83917806bbde320786a8fb61efebee

Hugging Face stats

6,765,653 downloads on Hugging Face · 5,100 likes · pipeline: text-generation · library: transformers

transformerssafetensorsgpt_osstext-generationvllmconversationaleval-resultsendpoints_compatible8-bitmxfp4

Weight Hashes

Weights file
3 shards (Merkle root) · 12.82 GB
SHA-256f9b21104c6edac7f6ecc4a526ac3cfadc6b4d864e04816bc4235df69775b15fa
SHA3-2564253ec14ee898a3523ac1db691c6d1246fb6b2cd3ef3e7c2bbcf35b757f02a2e
BLAKE38f569ae570631164690ad2f4f3e6b8fa565393afce2280dcc909cd1eca3f43e1
Config file
config.json · 1.76 KB
SHA-2563a2a26ded679375b7928ddeca59764df7cea83220c1961035f6d6e232659e9ce
SHA3-256e2c2399b3b9d574d2d39d4da99456fd0bdae700e94fd5603aa240a5dc9e067c1
BLAKE3768e499092fa4ced1cb5dc898a55135d996a908ad37a6f0fe62f75cdc59489ce
Tokenizer file
tokenizer.json · 26.58 MB
SHA-2560614fe83cadab421296e664e1f48f4261fa8fef6e03e63bb75c20f38e37d07d3
SHA3-256d03a0e559bd78c22acea8d7da22550cf57eafe09c635acb9de18fc08b5e5b7ba
BLAKE30596814de1ca4de2980968802fdb4340e84e39f8ffa67467648e8e606680acb1

Verify it yourself

Download the file(s) from the source below, then hash them locally (sha256sum <file> on Linux/macOS, certutil -hashfile <file> SHA256 on Windows) and compare the output against the hash(es) shown above.

Anomalies Flagged

  • filename declares 2 shards (of-00002) but found 3 files matching the pattern: ['model-00000-of-00002.safetensors', 'model-00001-of-00002.safetensors', 'model-00002-of-00002.safetensors'] -- verify manually