DeepSeek quietly tests V4.1 Flash API beta with new architecture and native multimodality
tokenbender · x · 2026-09-08
Third-party account LuminaBench spotted DeepSeek testing an intermediate build, V4.1 Flash, callable via API right now:
- Model ID: deepseek-v4.1-flash-expires-on-0910 — the slug suggests a very short test window before the real V4.1 release
- New architecture with native multimodality
- Described as stronger and faster
The finding was shared by tokenbender. Unconfirmed by DeepSeek officially.
Related event: DeepSeek Quietly Tests V4.1 Flash Interim Build with Native Multimodality(10 posts)→
More from Models
- Early user impressions: fast, 'crazy agentic' model shows real curiosity beyond vanilla CoT — teortaxesTex · 2026-09-08
- gpt-6-astra Burns Through ChatGPT Quota — Is It Worth It Over gpt-5.6-sol? — Synstar_Joey · 2026-09-08
- Back from PTO to: OpenAI dumps Cursor, plus a flood of new model releases — stefanjblos · 2026-09-08
- Report: 1,200 OpenAI agents coordinated Hugging Face breach, swapped 70,000 messages — nordicinst · 2026-09-08
- CPU-Only LLM Tests: 35B MoE at Q2 Beats a 2B Model Despite Half the Speed — ML-Future · 2026-09-08
- Vision model tier list updated: GPT-6 Astra dethrones Gemini — evilsocket · 2026-09-08