Pass IndexThe State of AISign in

DeepSeek-R1 (0528)

The DeepSeek R1 model has undergone a minor version upgrade, with the current version being DeepSeek-R1-0528.

text → text · made by DeepSeek

$0.2$0.6per Mtok in / out
Lambda · 12 sellers

Sold by 17 ways

SellerLaneRate
Google Vertex AI 0 ways$0.68per Mtok in$2.7per Mtok out
Google Vertex AI cloudnot on the seller's list todaybatch$0.68per Mtok in$2.7per Mtok outcloud.google.com · read 2026-08-24
Google Vertex AI 0 ways$1.35per Mtok in$5.4per Mtok out
Google Vertex AI cloudnot on the seller's list todaystandard$1.35per Mtok in$5.4per Mtok outcloud.google.com · read 2026-08-24
Crusoe cloudstandard$3per Mtok in$7per Mtok outraw.githubusercontent.com · read 2026-09-16
Lambda aggregatorstandard$0.2per Mtok in$0.6per Mtok outraw.githubusercontent.com · read 2026-09-16
Hyperbolic aggregatorstandard$0.25per Mtok in$0.25per Mtok outraw.githubusercontent.com · read 2026-09-16
DeepInfra aggregatorstandard$0.5per Mtok in$2.15per Mtok out$0.35per Mtok cachedapi.deepinfra.com · read 2026-08-25
Nous Research aggregatorstandard$0.5per Mtok in$2.15per Mtok out$0.35per Mtok cachedinference-api.nousresearch.com · read 2026-09-12
OpenRouter 4 ways$0.5per Mtok in$2.15per Mtok out$0.35per Mtok cached
OpenRouter aggregatorstandard$0.5per Mtok in$2.15per Mtok out$0.35per Mtok cachedopenrouter.ai · read 2026-08-24
OpenRouter aggregatorsiliconflow/fp8$0.5per Mtok in$2.18per Mtok outopenrouter.ai · read 2026-08-25
OpenRouter aggregatorstreamlake$0.57per Mtok in$2.29per Mtok outopenrouter.ai · read 2026-08-25
OpenRouter aggregatornovita/fp8$0.7per Mtok in$2.5per Mtok out$0.35per Mtok cachedopenrouter.ai · read 2026-08-25
ElectronHub aggregatorstandard$0.7per Mtok in$2.5per Mtok outapi.electronhub.ai · read 2026-09-16
Novita AI aggregatorstandard$0.7per Mtok in$2.5per Mtok out$0.35per Mtok cachedapi.novita.ai · read 2026-08-25
Fireworks AI aggregatorstandard$3per Mtok in$8per Mtok outraw.githubusercontent.com · read 2026-09-16
Hugging Face Inference Providers aggregatorstandard$3per Mtok in$5per Mtok outmodels.dev · read 2026-09-16
Together AI aggregatorstandard$3per Mtok in$7per Mtok outraw.githubusercontent.com · read 2026-09-16
PPIO aggregatorstandard$4per Mtok in$16per Mtok outapi.ppinfra.com · read 2026-09-10

Measured 28 standings

PlaceBoardMetricScore
5thof 28LiveCodeBench Leaderboard (default window 8/1/2024–5/1/2025, 454 problems)Pass@173.1
9thof 108Math level 5 — Epoch AImean_score96.639
14thof 49Lech mazur writing — Epoch AIMean score8.19
16thof 69Aider polyglot coding leaderboardPercent correct71.4
19thof 72Aider polyglot — Epoch AIPercent correct71.4
21stof 61AIME 2025Accuracy (± 95% CI)89.17
29thof 50Metr time horizons — Epoch AIaverage_score0.538
41stof 41Deepresearchbench — Epoch AIAverage score0.351
58thof 112Ale bench — Epoch AIPerformance804.12
71stof 102Simplebench — Epoch AIScore (AVG@5)0.408
80thof 152Design Arena · dataviz (models)Elo1197
93rdof 147Design Arena · 3d (models)Elo1139
93rdof 109Design Arena · svg (models)Elo1063
103rdof 170Weirdml — Epoch AIAccuracy41.63
104thof 162Design Arena · website (models)Elo1162
107thof 156Design Arena · codecategories (models)Elo1154
107thof 150Design Arena · uicomponent (models)Elo1118
110thof 155Design Arena · gamedev (models)Elo1120
117thof 402LMArena · TextScore (Elo)1421
143rdof 313GPQA Diamondmean_score76.326
147thof 290OTIS Mock AIME 2024-2025mean_score66.389
174thof 205ARC-AGI-1 (Public Eval)Score (%)26.98
177thof 216ARC-AGI-2 (Public Eval)Score (%)0.29
190thof 230ARC-AGI-1 (Semi-Private)Score (%)21.21
193rdof 227Arc agi 2 — Epoch AIScore1.12
197thof 232ARC-AGI-2 (Semi-Private)Score (%)1.12
202ndof 244Arc agi — Epoch AIScore21.21
411thof 656Epoch capabilities index — Epoch AIECI Score141.32

About

DeepSeek-R1 (0528) — a text model from DeepSeek, sold by 12 companies from $0.2 in and $0.6 out per million tokens, placed 9th of 108 on Math level 5 — Epoch AI.

It takes text and returns text, with a context window of 163,840 tokens. It was published in May 2025, trained on material up to July 2024. Its sellers say it can reason step by step and call a tool. The catalogue files it under reasoning, code and agents. Twelve companies sell it. The cheapest is $0.2 in and $0.6 out per million tokens at Lambda. Beside the standard rate there are regional and separately routed lanes. It has been measured on 28 boards, and stands best at 9th of 108 on Math level 5 — Epoch AI.

Every current figure

Maker
DeepSeek
Register
model
Takes
text
Returns
text
Context
163,840 tokens
Published
May 2025
Knowledge to
July 2024
Parameters
685 billion
Licence
not read
Sellers
12
Maker's own price
not read
Price
$0.2 in and $0.6 out per million tokens — Lambda
Boards
28
Best place
9th of 108 — Math level 5 — Epoch AI

Known as 4 names

DeepSeek R1 (0528)DeepSeek-R1-0528deepseek-ai/DeepSeek-R1-0528deepseek/deepseek-r1-0528