VERIFIEDOct 142026
Is DeepSeek V3.1 Terminus Thinking state of the art?
No.Best current rank is #71 in Long context.
As of Oct 14, 2026, DeepSeek V3.1 Terminus Thinking from DeepSeek. Best current rank is #71 in Long context. Ranked #71 in Long context.
Why
The ruling, in plain English
Ranked #96 in General LLM.
Receipts
Evidence
| Source | Score | Date | Type | Receipt |
|---|---|---|---|---|
| LMArena Text (style-controlled, overall) deepseek-v3.1-terminus-thinking | 1416.2971339559667 1406.2–1426.4 | 2026-10-13 | INDEP. | snap_7d2d37f2ecb4 |
| LMArena Text (style-controlled, hard prompts) deepseek-v3.1-terminus-thinking | 1444.614633132315 1430.2–1459.0 | 2026-10-13 | INDEP. | snap_7d2d37f2ecb4 |
| LMArena Text (style-controlled, coding) deepseek-v3.1-terminus-thinking | 1459.1638852743074 1435.6–1482.7 | 2026-10-13 | INDEP. | snap_7d2d37f2ecb4 |
| LMArena Text (style-controlled, longer query) deepseek-v3.1-terminus-thinking | 1446.5607762679765 1424.2–1468.9 | 2026-10-13 | INDEP. | snap_7d2d37f2ecb4 |
Reign history
Where it sits in the succession
No closed or open crown interval includes this model.
Honest about the gaps
Where it is not the crown
| Category | This model | Why |
|---|---|---|
| General LLM | #96 in this category | Ranked #96 in General LLM. |
| Reasoning | #86 in this category | Ranked #86 in Reasoning. |
| Coding | Not enough data | #100 on LMArena Text (style-controlled, coding); missing Terminal-Bench 4.0 and SWE-bench Pro V2 HARD |
| Long context | #71 in this category | Ranked #71 in Long context. |