Is DeepSeek V4 Pro Preview state of the art?#66Best current rank: #66 in Long context, as of Nov 22, 2026.
As of Nov 22, 2026, DeepSeek V4 Pro Preview is ranked #66 in Long context (since Nov 22, 2026; sources LMArena Text (style-controlled, longer query); methodology v1.0). Ranked #66 in Long context.
Facts and scores
DeepSeek V4 Pro Preview is an open-weights (MIT) text model from DeepSeek; no public parameter count and no public price.
How it scores
- On LMArena Text (style-controlled, math) it scores 1465 (CI 1453–1477), #63 of 231 (top 28%); 70 behind Gemini 4 Argon; level with Qwen3 7 Plus above and Grok 4.20 Beta is 1 below.
- On LMArena Text (style-controlled, overall) it scores 1455 (CI 1451–1459), #82 of 257 (top 32%); 72 behind Gemini 4 Carbon; level with Qwen3 7 Plus above and level with DeepSeek V4 Pro High Preview below.
- On LMArena Text (style-controlled, longer query) it scores 1464 (CI 1459–1470), #82 of 248 (top 34%); 76 behind Gemini 4 Argon; Ernie 5.1 is 1 above and level with Qwen3 7 Plus below.
History
First result on isitsota: Nov 19, 2026. Best overall rank so far: #66 in Long context, first held Nov 22, 2026.
The ruling, in plain English
Ranked #67 in General LLM.
Evidence
| Source | Score | Date | Type | Receipt |
|---|---|---|---|---|
| LMArena Text (style-controlled, overall) DeepSeek-V4-Pro Preview (High) | 1455 1451–1459 | 2026-11-19 | INDEP. | snap_280a9b5ca493 |
| LMArena Text (style-controlled, hard prompts) DeepSeek-V4-Pro Preview (High) | 1473 1469–1478 | 2026-11-19 | INDEP. | snap_280a9b5ca493 |
| LMArena Text (style-controlled, math) DeepSeek-V4-Pro Preview (High) | 1465 1453–1477 | 2026-11-19 | INDEP. | snap_280a9b5ca493 |
| LMArena Text (style-controlled, coding) DeepSeek-V4-Pro Preview (High) | 1488 1482–1494 | 2026-11-19 | INDEP. | snap_280a9b5ca493 |
| LMArena WebDev DeepSeek-V4-Pro Preview (High) | 1464 1457–1470 | 2026-11-20 | INDEP. | snap_2f4a3ebc7e46 |
| LMArena Text (style-controlled, longer query) DeepSeek-V4-Pro Preview (High) | 1464 1459–1470 | 2026-11-19 | INDEP. | snap_280a9b5ca493 |
| LMArena Text factuality DeepSeek-V4-Pro Preview (High) | 1448 1444–1451 | 2026-11-19 | INDEP. | snap_275d95909246 |
Where it sits in the succession
Crown intervals for this model are still empty.
Where it is not the crown
| Category | This model | Rank | Leader |
|---|---|---|---|
| General LLM | #67 in this category | #67 | Gemini 4 Carbon |
| Reasoning | #72 in this category | #72 | Gemini 4 Carbon |
| Math | Not enough data | — | GPT-6 Astra |
| Coding | Not enough data | — | Claude Sonnet 5.5 |
| Coding › Web dev | #73 in this category | #73 | Claude Fable 5.2 |
| Long context | #66 in this category | #66 | Gemini 4 Argon |
| Factuality | #69 in this category | #69 | Gemini 4 Carbon |
Related pages
- General LLM
- Reasoning
- Math
- Coding
- Coding › Web dev
- Long context
- Factuality
- LMArena WebDev
- LMArena Text (style-controlled, overall)
- LMArena Text (style-controlled, coding)
- LMArena Text (style-controlled, math)
- LMArena Text (style-controlled, hard prompts)
- LMArena Text (style-controlled, longer query)
- LMArena Text factuality
- Lab: DeepSeek
Is DeepSeek V4 Pro Preview SOTA?
Best current rank: #66 in Long context, as of Nov 22, 2026.
What is DeepSeek V4 Pro Preview best at?
As of Nov 22, 2026, DeepSeek V4 Pro Preview is ranked #66 in Long context (since Nov 22, 2026; sources LMArena Text (style-controlled, longer query)).
How does DeepSeek V4 Pro Preview rank in Coding?
As of Nov 22, 2026, DeepSeek V4 Pro Preview is short of the evidence for a rank in Coding (since Nov 22, 2026; sources LMArena Text (style-controlled, coding); methodology v1.0).