diff --git a/CHANGELOG.md b/CHANGELOG.md
index d4a8cfd7b..0cb957438 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased](https://github.com/qutip/QuantumToolbox.jl/tree/main)
 
+- Introduce `QuantumToolbox.settings` and `auto_tidyup`. ([#460])
+
 ## [v0.31.0]
 Release date: 2025-05-03
 
@@ -221,3 +223,4 @@ Release date: 2024-11-13
 [#453]: https://github.com/qutip/QuantumToolbox.jl/issues/453
 [#455]: https://github.com/qutip/QuantumToolbox.jl/issues/455
 [#456]: https://github.com/qutip/QuantumToolbox.jl/issues/456
+[#460]: https://github.com/qutip/QuantumToolbox.jl/issues/460
diff --git a/docs/make.jl b/docs/make.jl
index 21b051f04..218cf9653 100644
--- a/docs/make.jl
+++ b/docs/make.jl
@@ -64,6 +64,7 @@ const PAGES = [
         "Hierarchical Equations of Motion" => "users_guide/HEOM.md",
         "Solving for Steady-State Solutions" => "users_guide/steadystate.md",
         "Two-time correlation functions" => "users_guide/two_time_corr_func.md",
+        "QuantumToolbox Settings" => "users_guide/settings.md",
         "Extensions" => [
             "users_guide/extensions/cuda.md",
             "users_guide/extensions/cairomakie.md",
diff --git a/docs/src/resources/api.md b/docs/src/resources/api.md
index b9df5dbd9..0358ceb1d 100644
--- a/docs/src/resources/api.md
+++ b/docs/src/resources/api.md
@@ -299,6 +299,7 @@ AbstractLinearMap
 ## [Utility functions](@id doc-API:Utility-functions)
 
 ```@docs
+QuantumToolbox.settings
 QuantumToolbox.versioninfo
 QuantumToolbox.about
 gaussian
diff --git a/docs/src/users_guide/settings.md b/docs/src/users_guide/settings.md
new file mode 100644
index 000000000..be8d914e9
--- /dev/null
+++ b/docs/src/users_guide/settings.md
@@ -0,0 +1,36 @@
+# [QuantumToolbox Settings](@id doc:QuantumToolbox-Settings)
+
+In this section, we introduce the default global settings used throughout the package and show how to modify them.
+
+All settings are stored in [`QuantumToolbox.settings`](@ref).
+
+!!! warning "Differences from QuTiP"
+    Due to the differences in programming languages, solving algorithms, and many other reasons, these global settings (including their default values and usage) may be very different from those in `Python QuTiP`.
+
+## List of settings
+
+Here, we list out each setting along with the specific functions that will use it.
+
+- `tidyup_tol::Float64 = 1e-14` : tolerance for [`tidyup`](@ref) and [`tidyup!`](@ref).
+- `auto_tidyup::Bool = true` : Automatically tidyup during the following situations:
+    * Solving for eigenstates, including [`eigenstates`](@ref), [`eigsolve`](@ref), and [`eigsolve_al`](@ref).
+- (to be announced)
+
+## Change default settings
+
+First, we can check the current [`QuantumToolbox.settings`](@ref):
+
+```@example settings
+using QuantumToolbox
+
+QuantumToolbox.settings
+```
+
+Next, one can overwrite the default settings by
+
+```@example settings
+QuantumToolbox.settings.tidyup_tol = 1e-10
+QuantumToolbox.settings.auto_tidyup = false
+
+QuantumToolbox.settings
+```
\ No newline at end of file
diff --git a/src/QuantumToolbox.jl b/src/QuantumToolbox.jl
index 409bcfa54..40bfc245f 100644
--- a/src/QuantumToolbox.jl
+++ b/src/QuantumToolbox.jl
@@ -75,6 +75,7 @@ export permute
 export cache_operator, iscached, isconstant
 
 # Utility
+include("settings.jl")
 include("utilities.jl")
 include("versioninfo.jl")
 include("progress_bar.jl")
diff --git a/src/qobj/arithmetic_and_attributes.jl b/src/qobj/arithmetic_and_attributes.jl
index 8eecc5acd..957ef786d 100644
--- a/src/qobj/arithmetic_and_attributes.jl
+++ b/src/qobj/arithmetic_and_attributes.jl
@@ -629,27 +629,28 @@ purity(ρ::QuantumObject{ObjType}) where {ObjType<:Union{Ket,Bra}} = sum(abs2, 
 purity(ρ::QuantumObject{Operator}) = real(tr(ρ.data^2))
 
 @doc raw"""
-    tidyup(A::QuantumObject, tol::Real=1e-14)
+    tidyup(A::QuantumObject, tol::Real=settings.tidyup_tol)
 
 Given a [`QuantumObject`](@ref) `A`, check the real and imaginary parts of each element separately. Remove the real or imaginary value if its absolute value is less than `tol`.
 """
-tidyup(A::QuantumObject, tol::T = 1e-14) where {T<:Real} = QuantumObject(tidyup(A.data, tol), A.type, A.dimensions)
-tidyup(A::AbstractArray, tol::T2 = 1e-14) where {T2<:Real} = tidyup!(copy(A), tol)
+tidyup(A::QuantumObject, tol::T = settings.tidyup_tol) where {T<:Real} =
+    QuantumObject(tidyup(A.data, tol), A.type, A.dimensions)
+tidyup(A::AbstractArray, tol::T2 = settings.tidyup_tol) where {T2<:Real} = tidyup!(copy(A), tol)
 
 @doc raw"""
-    tidyup!(A::QuantumObject, tol::Real=1e-14)
+    tidyup!(A::QuantumObject, tol::Real=settings.tidyup_tol)
 
 Given a [`QuantumObject`](@ref) `A`, check the real and imaginary parts of each element separately. Remove the real or imaginary value if its absolute value is less than `tol`.
 
 Note that this function is an in-place version of [`tidyup`](@ref).
 """
-tidyup!(A::QuantumObject, tol::T = 1e-14) where {T<:Real} = (tidyup!(A.data, tol); A)
-function tidyup!(A::AbstractSparseArray, tol::T2 = 1e-14) where {T2<:Real}
+tidyup!(A::QuantumObject, tol::T = settings.tidyup_tol) where {T<:Real} = (tidyup!(A.data, tol); A)
+function tidyup!(A::AbstractSparseArray, tol::T2 = settings.tidyup_tol) where {T2<:Real}
     tidyup!(nonzeros(A), tol) # tidyup A.nzval in-place (also support for CUDA sparse arrays)
     return dropzeros!(A)
 end
-tidyup!(A::AbstractArray{T}, tol::T2 = 1e-14) where {T<:Real,T2<:Real} = @. A = T(abs(A) > tol) * A
-tidyup!(A::AbstractArray{T}, tol::T2 = 1e-14) where {T,T2<:Real} =
+tidyup!(A::AbstractArray{T}, tol::T2 = settings.tidyup_tol) where {T<:Real,T2<:Real} = @. A = T(abs(A) > tol) * A
+tidyup!(A::AbstractArray{T}, tol::T2 = settings.tidyup_tol) where {T,T2<:Real} =
     @. A = T(abs(real(A)) > tol) * real(A) + 1im * T(abs(imag(A)) > tol) * imag(A)
 
 @doc raw"""
diff --git a/src/qobj/eigsolve.jl b/src/qobj/eigsolve.jl
index 1d8f256e3..856afa956 100644
--- a/src/qobj/eigsolve.jl
+++ b/src/qobj/eigsolve.jl
@@ -249,6 +249,7 @@ function _eigsolve(
     end
     mul!(cache1, Vₘ, M(Uₘ * VR))
     vecs = cache1[:, 1:k]
+    settings.auto_tidyup && tidyup!(vecs)
 
     return EigsolveResult(vals, vecs, type, dimensions, iter, numops, (iter < maxiter))
 end
@@ -349,7 +350,10 @@ function eigsolve(
         vals = @. (1 + sigma * res.values) / res.values
     end
 
-    return EigsolveResult(vals, res.vectors, res.type, res.dimensions, res.iter, res.numops, res.converged)
+    vecs = res.vectors
+    settings.auto_tidyup && tidyup!(vecs)
+
+    return EigsolveResult(vals, vecs, res.type, res.dimensions, res.iter, res.numops, res.converged)
 end
 
 @doc raw"""
@@ -430,6 +434,8 @@ function eigsolve_al(
         @. vecs[:, i] = vec * exp(-1im * angle(vec[1]))
     end
 
+    settings.auto_tidyup && tidyup!(vecs)
+
     return EigsolveResult(vals, vecs, res.type, res.dimensions, res.iter, res.numops, res.converged)
 end
 
@@ -457,11 +463,11 @@ values:
      2.8569700138728056 + 0.0im
 vectors:
 5×5 Matrix{ComplexF64}:
-  0.106101+0.0im  -0.471249-0.0im  …   0.471249-0.0im  0.106101-0.0im
- -0.303127-0.0im   0.638838+0.0im      0.638838+0.0im  0.303127-0.0im
-  0.537348+0.0im  -0.279149-0.0im      0.279149-0.0im  0.537348-0.0im
+  0.106101+0.0im  -0.471249-0.0im  …   0.471249+0.0im  0.106101+0.0im
+ -0.303127-0.0im   0.638838+0.0im      0.638838+0.0im  0.303127+0.0im
+  0.537348+0.0im  -0.279149-0.0im      0.279149+0.0im  0.537348+0.0im
  -0.638838-0.0im  -0.303127-0.0im     -0.303127-0.0im  0.638838+0.0im
-  0.447214+0.0im   0.447214+0.0im     -0.447214-0.0im  0.447214-0.0im
+  0.447214+0.0im   0.447214+0.0im     -0.447214-0.0im  0.447214+0.0im
 
 julia> expect(H, ψ[1]) ≈ E[1]
 true
@@ -473,6 +479,7 @@ function LinearAlgebra.eigen(A::QuantumObject{OpType}; kwargs...) where {OpType<
     # This fixes a type inference issue. But doesn't work for GPU arrays
     E::mat2vec(to_dense(MT)) = F.values
     U::to_dense(MT) = F.vectors
+    settings.auto_tidyup && tidyup!(U)
 
     return EigsolveResult(E, U, A.type, A.dimensions, 0, 0, true)
 end
diff --git a/src/settings.jl b/src/settings.jl
new file mode 100644
index 000000000..b9be9b0b2
--- /dev/null
+++ b/src/settings.jl
@@ -0,0 +1,36 @@
+Base.@kwdef mutable struct Settings
+    tidyup_tol::Float64 = 1e-14
+    auto_tidyup::Bool = true
+end
+
+function Base.show(io::IO, s::Settings)
+    println(io, "QuantumToolbox.jl Settings")
+    println(io, "--------------------------")
+    map(x -> println(io, "$x = ", getfield(s, x)), fieldnames(Settings))
+    return nothing
+end
+
+@doc raw"""
+    QuantumToolbox.settings 
+
+Contains all the default global settings of QuantumToolbox.jl.
+
+# List of settings
+
+- `tidyup_tol::Float64 = 1e-14` : tolerance for [`tidyup`](@ref) and [`tidyup!`](@ref).
+- `auto_tidyup::Bool = true` : Automatically tidyup.
+
+For detailed explanation of each settings, see our documentation [here](https://qutip.org/QuantumToolbox.jl/stable/users_guide/settings).
+
+# Change default settings
+
+One can overwrite the default global settings by
+
+```julia
+using QuantumToolbox
+
+QuantumToolbox.settings.tidyup_tol = 1e-10
+QuantumToolbox.settings.auto_tidyup = false
+```
+"""
+const settings = Settings()