DeepSeek R1
DeepSeek · Text & chat · Text · Deep reasoning · 128K context
DeepSeek — reasoning-first models that think before they answer.
DeepSeek's V3 is a capable, low-cost generalist, and R1 is a 'thinking' model that works through maths, logic and coding problems step by step. Use R1 when the answer has to be right, not just fluent.
Strengths
- Step-by-step reasoning (R1)
- Maths and code
- Very competitive pricing
Best for
- Logic and quantitative problems
- Debugging and algorithms
- Cheap bulk generation (V3)
Reach for another model when: Latency-sensitive chat (R1 thinks first).
How DeepSeek R1 works inside Kryotta
Pick DeepSeek R1 in the model picker and chat as usual — or leave the picker on Auto and Kryotta routes each prompt to the model that does the job best for the least cost, DeepSeek R1 included. Every answer shows which model produced it. Add it to a Compare run to see it against Claude 3.5 Sonnet and Claude 3 Haiku on the exact same prompt. Usage from every model counts against one pooled allowance, so your team gets one predictable bill.
Frequently asked
What is DeepSeek R1 best at?
DeepSeek — reasoning-first models that think before they answer. It shines at: Logic and quantitative problems, Debugging and algorithms, Cheap bulk generation (V3).
How do I use DeepSeek R1 in Kryotta?
Sign in, pick DeepSeek R1 from the model picker in Chat (or leave it on Auto and Kryotta routes prompts to it when it's the best value), or add it to a Compare run to see it next to other models on the same prompt.
Which plan includes DeepSeek R1?
DeepSeek R1 is available on Pro and Enterprise plans.
Can I compare DeepSeek R1 with other models?
Yes — the Compare arena runs the same prompt through up to several models at once and shows answers, speed and cost side by side.