DeepSeek-V3.1 at a glance
DeepSeek-V3.1 merged chat and reasoning into one hybrid model with 128K context, stronger tool use and agent skills, Anthropic API compatibility and open weights.
What changed with DeepSeek-V3.1
On August 21, 2025, DeepSeek released DeepSeek-V3.1, which it called its first step toward the agent era. One model now offers both a thinking and a non-thinking mode: in the API, deepseek-chat maps to non-thinking and deepseek-reasoner to thinking, both with a 128K context window. DeepSeek said V3.1 in thinking mode reaches answers faster than R1-0528.
Post-training improved tool use and multi-step agent tasks, with better SWE and Terminal-Bench results by DeepSeek's account. The API added support for the Anthropic API format and strict function calling in beta. The base model received 840 billion tokens of continued pretraining for longer context, and both V3.1 and V3.1-Base weights were published on Hugging Face. New API pricing took effect on September 5, 2025.
DeepSeek-V3.1 in numbers
Values are copied from the linked source on the date shown. They are not IA Compa measurements.
- Context windowas of · Source for Context window: api-docs.deepseek.com (opens in a new tab)
- 128K tokens
How it compares
Other models in the same category that we have covered, using each model's latest recorded value. (Models)
Where to compare DeepSeek-V3.1
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
General leaderboards
Why DeepSeek-V3.1 matters for businesses
For businesses, V3.1 simplified model choice to a single low-cost open model that can either answer quickly or reason, and Anthropic API compatibility made it easy to test inside existing agent tools.
Sources for this story
- DeepSeek-V3.1 Release (opens in a new tab)Official
DeepSeek API Docs · api-docs.deepseek.com
- deepseek-ai/DeepSeek-V3.1 (opens in a new tab)Official
Hugging Face · huggingface.co
- DeepSeek V3.1 model page (opens in a new tab)Benchmark
Artificial Analysis · artificialanalysis.ai
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #llm
- #deepseek
- #open-weights
- #reasoning
- #agents