# Is DeepSeek V4.1 Flash SOTA?

**Yes.** Yes on General LLM (best per dollar); Reasoning (best per dollar). Not the overall leader.

As of Oct 14, 2026, DeepSeek V4.1 Flash from DeepSeek. Yes on General LLM (best per dollar); Reasoning (best per dollar). Not the overall leader. On the cost-quality frontier for General LLM: nothing scores better for less than $0.24 per 1M blended tokens.

As of Oct 14, 2026. Methodology v1.0. Computed automatically.

- General LLM: RANKED #29 — Ranked #29 in General LLM.
- Reasoning: RANKED #27 — Ranked #27 in Reasoning.
- Math: INSUFFICIENT_DATA — #18 on LMArena Text (style-controlled, math); missing Humanity's Last Exam and FrontierMath
- Science: RANKED #36 — Last listed at #36 in Science on 2026-09-09. That board does not award a crown.
- Coding: INSUFFICIENT_DATA — #21 on LMArena Text (style-controlled, coding); missing Terminal-Bench 4.0 and SWE-bench Pro V2 HARD
- Coding › Web dev: RANKED #18 — Ranked #18 in Coding › Web dev.
- Long context: RANKED #31 — Ranked #31 in Long context.
- Vision: RANKED #21 — Ranked #21 in Vision.
- Factuality: RANKED #22 — Ranked #22 in Factuality.
- Abstraction (ARC): RANKED #8 — Last listed at #8 in Abstraction (ARC) on 2026-09-09. That board does not award a crown.

## Factuality

https://isitsota.com/c/factuality