DeepSeek V5 leak: 2T parameters, reportedly trained fully on Huawei Ascend chips
teortaxesTex · x · 2026-09-27
An unverified leak claims DeepSeek is preparing an imminent V5 launch: rumored at 2 trillion parameters (not 3T), described by founder Liang Wenfeng as the company's biggest bet yet. It would reportedly be the first DeepSeek model trained fully on Huawei Ascend chips instead of Nvidia — though needing roughly 4x more Ascend accelerators to hit the same training scale. The open-weight strategy is said to continue. A joked-about "independent corroboration" has surfaced from Nigeria. All rumors, unconfirmed.
More from Infra
- India's Sarvam launches Token Factory, hosting top open models on local infra — HarveenChadha · 2026-11-16
- Gemini 3.5 Flash now runs on air-gapped Google Distributed Cloud, offline up to 12 months — shashib · 2026-11-16
- Gemma.c: full Gemma 1/2/3 multimodal inference in a single C file, ~72 tok/s on CPU at INT8 — This-Purchase-3325 · 2026-11-16
- BIS Issues Interim Final Rule Addressing Remote Access to Advanced AI Chips — apples_jimmy · 2026-11-16
- Jailbroken PS4 runs a 9B open-source model at up to 43 tok/s by talking directly to its 2013-era GPU — AIFlow_ML · 2026-11-16
- Mobile AI Factory: agents auto-build on-device SLMs with 19x less GPU time — pmttyji · 2026-11-16