DeepSeek R1 (0528): benchmark scores, price and ranking

DeepSeek R1 (0528) from deepseek-ai ranks #33 of 743 LLMs on the LLMs Tiger aggregate benchmark score (53.6), based on 3 benchmark results captured up to 2026-10-13.

Organisation: deepseek-ai
Type: Open-weight / local model
Open weights: Yes
Context window: 163,840 tokens
Cheapest API price: $0.400 in / $0.400 out per million tokens (via hyperbolic, observed 2026-10-13T20:47Z)
Aggregate rank: #33 of 743 (score 53.6)

Relative to other models, DeepSeek R1 (0528) is strongest on MMLU-Pro (#14 of 136) and weakest on GPQA Diamond (#113 of 275).

Across the 3 benchmarks where at least ten models are scored, DeepSeek R1 (0528) is in the top 10% on 0 and in the top half on 3.

Its cheapest listed input price, $0.400 per million tokens, is lower than that of 53% of the 427 models with a listed price.

DeepSeek R1 (0528) benchmark results

DeepSeek R1 (0528) benchmark scores, ranked against all models measured on each benchmark
Benchmark	Score	Rank	Source	Captured
MMLU-Pro	85.0%	#14 of 136	HuggingFace MMLU-Pro leaderboard	2026-10-13
Aider Polyglot	71.4%	#7 of 49	Aider polyglot leaderboard	2026-10-13
GPQA Diamond	78.8%	#113 of 275	HuggingFace GPQA Diamond leaderboard	2026-10-13

About DeepSeek R1 (0528)

DeepSeek R1 (0528) is an open-weight model that can run on your own hardware from deepseek-ai. It accepts up to 163,840 tokens of input at once. The cheapest API listing we track charges $0.400 per million input tokens and $0.400 per million output tokens, through hyperbolic. We hold 3 benchmark results for it across 2 areas, and none is in the top tenth of its board.

Where DeepSeek R1 (0528) stands

On our aggregate score, DeepSeek R1 (0528) is #33 of 743 ranked models, in the top fifty. Its score is 53.6; the scale is centred on 50, so it sits 3.6 points above the typical model. That beats 96% of the ranked field. Among the 7 ranked deepseek-ai models it is #1. The next one down is DeepSeek-V2.5 at 50.9.

DeepSeek R1 (0528), benchmark by benchmark

MMLU-Pro puts DeepSeek R1 (0528) at 85.0%, #14 of 136. That beats 90% of the models tested, but 3.1 points short of Darwin-180B-RSI (88.1%). The results either side are Qwen3.6 35B-A3B (85.2%) and glm-4.5 (84.6%). It covers 57-subject knowledge + reasoning. The board median is 55.3%, so DeepSeek R1 (0528) is 29.7 points above it. Source: HuggingFace MMLU-Pro leaderboard, captured 2026-10-13.

On Aider Polyglot, DeepSeek R1 (0528) reaches 71.4%. It ranks #7 of 49, ahead of 86% of the field, and 11.7 points behind gemini-2.5-pro-preview-06-05 at 83.1%. The results either side are claude-opus-4-20250514 (72.0%) and claude-3-7-sonnet-20250219 (64.9%). It covers edits across 6 languages (225 exercises). The board median is 44.9%, so DeepSeek R1 (0528) is 26.5 points above it. Source: Aider polyglot leaderboard, captured 2026-10-13.

DeepSeek R1 (0528) manages 78.8% on GPQA Diamond, good for #113 of 275. GPT-6 Astra (max) leads at 95.8%, 17.0 points further up, and 59% of models sit below. The results either side are Claude Sonnet 4.6 (max) (78.8%) and claude-sonnet-4-20250514_32K (78.3%). It covers expert-level science questions. The board median is 71.7%, so DeepSeek R1 (0528) is 7.1 points above it. Source: HuggingFace GPQA Diamond leaderboard, captured 2026-10-13.

Strengths and weaknesses

Reasoning, maths and science: solid, ahead of 74% of the field on average across 2 benchmarks. Its best board here is MMLU-Pro (#14 of 136) and its weakest is GPQA Diamond (#113 of 275).

Price and value

The median input price across the 427 priced models is $0.500 per million tokens, so DeepSeek R1 (0528) costs less than the typical listing. Models that score within 1.5 points or better and cost less: DeepSeek V4 Flash 0731 (max) ($0.016, 53.1), Qwen3-Coder-30B-A3B-Instruct ($0.070, 52.7), GLM-5.3 (max) ($0.079, 53.8). Similar scores cost more on mimo-v2.6-pro ($0.435) and Kimi K2.5 (Fireworks) ($0.450).

Reading these numbers

The 3 scores on this page come from 3 sources: Aider polyglot leaderboard (1), HuggingFace GPQA Diamond leaderboard (1), HuggingFace MMLU-Pro leaderboard (1). 1 of them come from an official leaderboard and 2 from datasets or third-party runs. With 3 results the aggregate rank can move a lot when one new score arrives.