In the news
Inference engineering for DeepSeek V4 Pro 0813
Baseten · Published · 3 min read
In 30 seconds
- What happened
- DeepSeek V4 Pro 0813, a 1.7 trillion parameter open-source model, was released today under MIT license with production API support.
- Why it matters
- Matters for engineers building AI applications who need frontier model quality at lower cost than closed alternatives without licensing restrictions.
- Watch out
- The model includes a native speculator for speculative decoding, but custom speculators trained on your specific workload would likely improve throughput further.
Listen to this summary
- inference
- deepseek
The patterns behind this
Each one covers how the technique works, when it earns its cost, and where it breaks.
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.