Model roadmap hints at a late-June release and a possible 150B run
teortaxesTex · x · 2026-07-23
The image quotes a discussion about the next release cadence and the company's model roadmap.
Key points:
- The current GCV4 release is described as still rough.
- The next version may arrive around late June, with a comfortable cadence of one release every 2–3 months.
- At the 50B activated-parameter scale, the speaker thinks the model will not differ much from the current open-source wave.
- They say the gap to a larger unpublished model is still large and would likely require something around 150B.
- Based on current progress, training 150B might start by year-end or at least by next April.
The accompanying text says V4 GA is being delayed, that it was wanted in June, and that Wenfeng hoped to start training a model with 150B active parameters by late 2026 to April 2027.
Related event: DeepSeek Founder's Investor Call Reveals AGI Roadmap and Compute Plans(37 posts)→
More from Models
- DeepSeek V4.1 Flash Hits 98% of GPT-6 Astra's Score at 1.4% of the Cost in Third-Party Benchmark — ayushtweetshere · 2026-09-11
- TheZvi Polls: Has Your Coding Model Choice Changed Since Fable 5.1 and Astra? — TheZvi · 2026-09-11
- antirez Weighs In on Anthropic Banning Minors From Using Claude — antirez · 2026-09-11
- Engram's random reads don't suit SSDs; CPU-memory over NVLink could serve all 72 GPUs — bookwormengr · 2026-09-11
- Fully local voice assistant on an RTX 3060 replicates the GPT Live demo in 6.5 minutes — liampetti · 2026-09-11
- Meta's Muse Agent has built-in invite code logic, hinting at free-usage expansion — testingcatalog · 2026-09-11