Pass IndexThe State of AISign in

Grok 3 Mini

Grok 3 Mini is a lightweight model that thinks before responding: fast, smart, and great for logic-based tasks that do not require deep domain knowledge, with the raw thinking traces accessible.

text → text · made by xAI

$0.25$1.27per Mtok in / out
Microsoft Azure AI · 3 sellers

Sold by 5 ways

SellerLaneRate
Microsoft Azure AI cloudstandard$0.25per Mtok in$1.27per Mtok outraw.githubusercontent.com · read 2026-09-16
Oracle Cloud Infrastructure cloudstandard$0.3per Mtok in$0.5per Mtok outraw.githubusercontent.com · read 2026-09-16
Requesty aggregatornot on the seller's list todayxai high$0.27per Mtok in$0.45per Mtok out$0.27per Mtok cachedrouter.requesty.ai · read 2026-08-25
Requesty aggregatornot on the seller's list todayxai low$0.27per Mtok in$0.45per Mtok out$0.27per Mtok cachedrouter.requesty.ai · read 2026-08-25
Requesty aggregatorxai$0.3per Mtok in$0.5per Mtok out$0.3per Mtok cachedrouter.requesty.ai · read 2026-08-28

Measured 17 standings

PlaceBoardMetricScore
11thof 28LiveCodeBench Leaderboard (default window 8/1/2024–5/1/2025, 454 problems)Pass@166.7
14thof 42Fictionlivebench — Epoch AI120k token score (medium)0.656
21stof 108Math level 5 — Epoch AImean_score (low)90.937
36thof 49Lech mazur writing — Epoch AIMean score (low)7.35
40thof 72Aider polyglot — Epoch AIPercent correct (high)49.3
67thof 101FrontierMath (Tiers 1-3)mean_score (high)5.862
99thof 170Weirdml — Epoch AIAccuracy (high)42.58
115thof 290OTIS Mock AIME 2024-2025mean_score (high)77.778
145thof 313GPQA Diamondmean_score (low)76.263
183rdof 205ARC-AGI-1 (Public Eval)Score (%) (low)17.62
190thof 402LMArena · TextScore (Elo) (high)1364
198thof 230ARC-AGI-1 (Semi-Private)Score (%) (low)16.5
198thof 216ARC-AGI-2 (Public Eval)Score (%) (low)0
207thof 227Arc agi 2 — Epoch AIScore (low)0.42
210thof 244Arc agi — Epoch AIScore (low)16.5
212thof 232ARC-AGI-2 (Semi-Private)Score (%) (low)0.42
430thof 656Epoch capabilities index — Epoch AIECI Score (high)140.37

About

Grok 3 Mini — a text model from xAI, sold by 3 companies from $0.25 in and $1.27 out per million tokens, placed 21st of 108 on Math level 5 — Epoch AI.

It takes text and returns text, with a context window of 131,072 tokens. It was published in June 2024, trained on material up to June 2024. Its sellers say it can call a tool. The catalogue files it under evaluate, reasoning and code. Three companies sell it. The cheapest is $0.25 in and $1.27 out per million tokens at Microsoft Azure AI. Beside the standard rate there is a separately routed lane. It has been measured on 17 boards, and stands best at 21st of 108 on Math level 5 — Epoch AI.

Every current figure

Maker
xAI
Register
model
Takes
text
Returns
text
Context
131,072 tokens
Published
June 2024
Knowledge to
June 2024
Licence
not read
Sellers
3
Maker's own price
not read
Price
$0.25 in and $1.27 out per million tokens — Microsoft Azure AI
Boards
17
Best place
21st of 108 — Math level 5 — Epoch AI

Known as 5 names

Grok-3-Mini (High)Grok-3 Minixai/grok-3-minixai/grok-3-mini:highxai/grok-3-mini:low