{"type":"post","id":"ps5lm-llama-cpp-on-a-jailbroken-ps5-run-aba8a6","slug":"ps5lm-llama-cpp-on-a-jailbroken-ps5-run-aba8a6","title":"PS5LM: llama.cpp on a jailbroken PS5: run any GGUF model natively on the console, starting with Qwen 3.8","description":"**Thesis:** PS5LM makes a console a constrained local-inference target by porting a general inference runtime, rather than implementing kernels for a single model.","retrieval_nugget":"**Thesis:** PS5LM makes a console a constrained local-inference target by porting a general inference runtime, rather than implementing kernels for a single model.","published_at":"2026-10-14","updated_at":"2026-10-14","record_date":"2026-10-14","date_kind":"published_at","topics":["inference"],"source_urls":["https://github.com/cobanov/PS5LM"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:ps5lm-llama-cpp-on-a-jailbroken-ps5-run-aba8a6","status":"published","visuals":[],"editorial_provenance":{"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:47f3bebfa017d58238be5304cf9183be0d49635b9fe249ab93d55a6acb2a4774","reviewed_at":"2026-10-14T22:03:47.409Z","source_evidence_count":1,"verified_claim_count":3,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":3,"implication_count":2,"watch_condition_count":1,"related_record_count":0},"analysis":{"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"PS5LM makes a console a constrained local-inference target by porting a general inference runtime, rather than implementing kernels for a single model.","observed_facts":[{"text":"The repository says PS5LM ports llama.cpp to a jailbroken PS5 and v0.1 provides a browser-based model library.","source_urls":["https://github.com/cobanov/PS5LM"]},{"text":"The repository says the current payload can keep about 1.6 GB of model in memory and lists Qwen3.5 0.8B and 2B.","source_urls":["https://github.com/cobanov/PS5LM"]},{"text":"The repository says Qwen 3.8 27B has not yet been run on a PS5 and that larger models need a native app.","source_urls":["https://github.com/cobanov/PS5LM"]}],"mechanism":"It ports llama.cpp to a jailbroken PS5 and exposes a browser-based model library; the current payload’s memory ceiling limits the models it can run.","why_now":"The repository has released v0.1 and documents a runnable path on specified firmware, shifting the project from a social claim to a checkable implementation.","implications":["Porting a general runtime broadens the candidate model architectures and quantizations, but available console memory—not the breadth of llama.cpp support—sets the immediate practical boundary.","The project requires a jailbroken console, limiting relevance to experimenters rather than ordinary PS5 owners."],"evidence_boundary":"The repository says the payload can keep about 1.6 GB of model in memory, lists Qwen3.5 0.8B and 2B, and says Qwen 3.8 27B has not yet been run on PS5; it does not substantiate a 27B console run. The setup requires a jailbroken console on firmware 7.00–13.60, so this is not ordinary PS5 functionality.","watch_conditions":["Whether a native app removes the current payload’s model-memory ceiling and enables verified runs of larger GGUF models is the next meaningful test."],"related_records":[],"new_branch_reason":"No prior New Runtime record is supplied for this project; this entry starts a distinct record on console-based local inference and its runtime and memory constraints."},"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/ps5lm-llama-cpp-on-a-jailbroken-ps5-run-aba8a6/","markdown":"https://newruntime.com/posts/ps5lm-llama-cpp-on-a-jailbroken-ps5-run-aba8a6.md","json":"https://newruntime.com/posts/ps5lm-llama-cpp-on-a-jailbroken-ps5-run-aba8a6.json"},"related_patterns":["harness-architecture-outlives-model-choice"],"next_reads":[{"type":"topic","path":"/topics/inference/","url":"https://newruntime.com/topics/inference/","title":"Inference","reason":"Explore the inference topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/inference.md","json_url":"https://newruntime.com/topics/inference.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/source-ledger.json","url":"https://newruntime.com/source-ledger.json","title":"Source Ledger","reason":"Inspect the public source evidence graph.","media_type":"application/json"}]}