Skip to content
aiqbench

DeepSeek · model passport

DeepSeek: R1 0528

deepseek-r1-0528
  • Open weights
  • MIT licence
  • 684.5B params
  • 163,840 context
  • Released 2025-05-28(curated)
12/27axes covered
Overall · LMArena
1,421#118 of 396
Best percentile
90.6MMLU-Pro (Artificial Analysis)
Cohort standing
58.5median 50.4

What the data says

Benchmarked on 12 of 27 tracked axes. Every figure below is the model's own published result on that board.

Strongest axes

  • 90.6 pct — MMLU-Pro (Artificial Analysis) (84.9%)
  • 86.6 pct — LiveCodeBench (Artificial Analysis) (77.0%)

Uneven: 71.5 percentile points between its best and worst axis — 90.6 pct on MMLU-Pro (Artificial Analysis) versus 19.1 pct on ARC-AGI-2.

  • MMLU-Pro (Artificial Analysis)90.6 pct84.9%Verify →
  • LiveCodeBench (Artificial Analysis)86.6 pct77.0%Verify →
  • LiveBench80.4 pct69.4%Verify →
  • Overall70.5 pct1,421Verify →
  • Coding69.6 pct1,464Verify →
  • Hard prompts68.2 pct1,433Verify →
  • Instruction following63.4 pct1,391Verify →
  • GPQA Diamond57 pct76.3%Verify →
  • AIME (math)47 pct66.4%Verify →
  • IFBench (Artificial Analysis)40.1 pct39.6%Verify →
  • SimpleBench34.5 pct40.8%Verify →
  • ARC-AGI-219.1 pct1.1%Verify →
  • WebDev ArenaNot benchmarked
  • Vision ArenaNot benchmarked
  • Factuality ArenaNot benchmarked
  • Search ArenaNot benchmarked
  • Document ArenaNot benchmarked
  • Agent ArenaNot benchmarked
  • FrontierMathNot benchmarked
  • SimpleQA VerifiedNot benchmarked
  • Humanity's Last ExamNot benchmarked
  • BFCL v4 (function calling)Not benchmarked
  • SWE-bench Verified (live board)Not benchmarked
  • Terminal-Bench 4.0Not benchmarked
  • HHEM Hallucination (Vectara)Not benchmarked
  • SWE-bench VerifiedNot benchmarked
  • Terminal-BenchNot benchmarked

Percentiles are the model's standing within each board's own population. Axes marked "not benchmarked" have no published result for this model — that is an absence of data, not a low score.

How to run it

Run it yourself

Repositorydeepseek-ai/DeepSeek-R1-0528
Parameters684.5B
ArchitectureDeepseekV3ForCausalLM
LicenceMIT

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "deepseek-ai/DeepSeek-R1-0528"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

Ollama

We don't sync Ollama tags, so we won't print a command we can't verify — a wrong one would pull the wrong model. Search the Ollama library →

Unsloth fine-tuning recipes are model-specific and we don't derive them here. Unsloth docs →

Use it hosted

API model IDdeepseek/deepseek-r1-0528
Context length163,840 tokens
Max output32,768 tokens
Input price$0.50 per 1M tokens
Cached input price$0.35 per 1M tokens
Output price$2.15 per 1M tokens

API parameters this endpoint supports

Frequency penaltyInclude reasoningLogit biasLogprobsMax tokensMin-pPresence penaltyReasoningRepetition penaltyResponse formatSeedStop sequencesStructured outputsTemperatureTool choiceTool callingTop-kTop logprobsTop-p

Example request

curl https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek/deepseek-r1-0528","messages":[{"role":"user","content":"Hello"}]}'

View on OpenRouter →

Identifiers, limits and prices are synced from the upstream source shown above. Any field we haven't synced is left out entirely rather than guessed.

  • deepseek-r1-0528 stands at 58.5 (raw mean 60.6) — measured on 12 of 27 axes.

    standingtemplateVerify →

  • deepseek-r1-0528 stands at 58.5 versus a cohort median of 50.4 across 145 models released within 180 days — above the peer median.

    cohorttemplateVerify →

Every claim is a computed fact from the current board data, phrased by a low-temperature model call and rejected by a validation gate if any number is not in the source row. How the leaderboards work →