DeepSeek
ReasoningPriced via OpenRouter

R1 0528

text->text

May 2025 R1 refresh. Improves long-form reasoning consistency and reduces hallucinations vs original R1.

R1 0528 is a reasoning AI model from DeepSeek. It costs $0.5000 per million input tokens and $2.15 per million output tokens (blended $0.9950/M), with a 163,840-token context window.

Profile inherited from upstream DeepSeek-R1-0528 ↗ — this is a hosted variant of the same open-weights model.

RatesUSD per million tokens. Blend is 70% input, 30% output, our calculation.

INPUT
$0.5000/M
per million input tokens
OUTPUT
$2.15/M
per million output tokens
BLENDED 70/30
$0.9950/M
unchanged since 3 May
CONTEXT
163,840
tokens

From OpenRouter's listing, read on 17 Nov 2026.

What it is good at
  • Open-weights reasoning
  • Improved over original R1
  • Cheap
  • Visible thinking
Typical use cases
  • Reasoning agents
  • Math/STEM tutoring
  • Migration from R1

Benchmarks

vs. best public score
Scores inherited from DeepSeek-R1-0528 — this is a hosted variant of the same open-weights model, so the underlying benchmark scores are identical.
Graduate-level science questions, "Google-proof". Source: DeepSeek (self-reported), GPQA-Diamond (Pass@1), read 2026-11-10.
Real GitHub issues solved end-to-end. Source: DeepSeek (self-reported), SWE Verified (Resolved), read 2026-11-10.
LMArena Elo1422 Elo
Crowd-sourced head-to-head preference Elo rating. Source: LMArena (independent), LMArena Text Arena score, read 2026-11-10.
Hand-curated from each provider's published reports and public leaderboards. Methodology varies across sources, treat as directional rather than authoritative.

Also reported by the provider

Results on benchmarks outside the set we compare across models. They are listed for completeness and are deliberately excluded from the bars above and from every ranking on this site, because a score is only comparable against the same benchmark variant. Why benchmark scores do not compare →

  • MMLU-Redux (EM)93.4%self-reported by DeepSeek · 2025-05 · R1-0528 model card. MMLU-Redux is a re-annotated MMLU subset, not the original test. Temperature 0.6, top-p 0.95, 16 responses per query, up to 64K tokens. The card also gives MMLU-Pro (EM) 85.0. Not comparable with the MMLU column, so shown here and never ranked.

More from DeepSeek

See all 11 →

Frequently asked questions

How much does R1 0528 cost?

R1 0528 costs $0.5000 per million input tokens and $2.15 per million output tokens, for a blended reference rate of $0.9950 per million tokens.

What is R1 0528's context window?

R1 0528 supports up to 163,840 tokens of context in a single request.

What is R1 0528 best for?

R1 0528 is well suited to Open-weights reasoning, Improved over original R1 and Cheap.

Who makes R1 0528?

R1 0528 is developed and served by DeepSeek. It was released in May 2025.

Terms used on this page