DeepSeek R1

Pricing — 1 offering(s)

Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.

Capability profile

Operator guidance

Route here when reasoning depth is the primary requirement and o1/o3 cost is prohibitive. Open weights mean it can be self-hosted or accessed via third-party hosts (Together AI, OpenRouter) at significantly lower cost than native DeepSeek API. Avoid for latency-sensitive or simple workloads — the reasoning overhead adds substantial time and token cost.

Use cases

• Complex multi-step reasoning, math, and science problems
• Advanced code generation and algorithm design
• Open-weights reasoning deployment on own infrastructure

Limitations

• The deepseek-reasoner API alias stopped naming this model back on 2025-08-21 (DeepSeek-V3.1's hybrid thinking/non-thinking unification); it now routes to deepseek-v4-flash's thinking mode and retires entirely 2026-07-24 — use open-weights hosts (R1 / R1-0528) for longevity
• Reasoning tokens can 3–5× the effective cost on complex problems
• Chinese lab; data-residency or compliance constraints may apply
• Capability ratings are qualitative, not a single cited benchmark run

Citations

• DeepSeek-R1: Incentivizing Reasoning Capability in LLMs (arXiv)
• DeepSeek API pricing