DeepSeek V4 Pro
AI models review
As of Jun 25, 2026, DeepSeek V4 Pro is rated 1458 Elo on LM Arena (26,928 Arena votes).
Benchmark and price facts
- Vendor
- DeepSeek
- License
- open
- LM Arena text
- 1458 Elo (#36 of 50)
- LM Arena coding
- 1458 Elo (#24)
- Price (input)
- $0.44 / 1M tokens
- Price (output)
- $0.87 / 1M tokens
- Context window
- 1,048,576 tokens
- Human votes
- 26,928
Elo and votes from LM Arena; intelligence, coding and speed indices from Artificial Analysis; price and context window from OpenRouter. Real fetched values, attributed, never blended.
Is DeepSeek V4 Pro worth it in 2026?
DeepSeek V4 Pro is a mid-tier open model sitting at rank 36 of 50 in general text (Elo 1458) and rank 24 in coding (Elo 1458), based on a solid 26928 human votes. Its very low price and massive 1M token context window are its clearest strengths.
High-volume or budget-conscious workloads, long-document processing, and coding tasks where cost matters more than top-tier quality.
General text quality ranks in the bottom half of the 50-model field, so users needing best-in-class chat or reasoning should look elsewhere.
AI-generated editorial synthesis of the source scores and sampled review themes, not a hands-on test. Reviewer experiences and allegations are not independently verified facts.
Rated near DeepSeek V4 Pro
Closest scores in ai models, on the same source.