diff --git a/lib/jax_finufft_cpu.cc b/lib/jax_finufft_cpu.cc
index ddd4563..10cf204 100644
--- a/lib/jax_finufft_cpu.cc
+++ b/lib/jax_finufft_cpu.cc
@@ -4,9 +4,11 @@
 
 #include "jax_finufft_cpu.h"
 
+#include "jax_finufft_cpu_opts.h"
 #include "pybind11_kernel_helpers.h"
 
 using namespace jax_finufft;
+namespace py = pybind11;
 
 namespace {
 
@@ -92,6 +94,48 @@ PYBIND11_MODULE(jax_finufft_cpu, m) {
   m.def("registrations", &Registrations);
   m.def("build_descriptorf", &build_descriptor<float>);
   m.def("build_descriptor", &build_descriptor<double>);
+
+  py::class_<jax_finufft_opts> opts(m, "Opts");
+
+  py::enum_<jax_finufft_opts::DebugLevel>(opts, "DebugLevel")
+      .value("Silent", jax_finufft_opts::DebugLevel::Silent)
+      .value("Vebose", jax_finufft_opts::DebugLevel::Verbose)
+      .value("Noisy", jax_finufft_opts::DebugLevel::Noisy)
+      .export_values();
+
+  py::enum_<jax_finufft_opts::FftwFlags>(opts, "FftwFlags")
+      .value("Estimate", jax_finufft_opts::FftwFlags::Estimate)
+      .value("Measure", jax_finufft_opts::FftwFlags::Measure)
+      .value("Patient", jax_finufft_opts::FftwFlags::Patient)
+      .value("Exhaustive", jax_finufft_opts::FftwFlags::Exhaustive)
+      .value("WisdomOnly", jax_finufft_opts::FftwFlags::WisdomOnly)
+      .export_values();
+
+  py::enum_<jax_finufft_opts::SpreadSort>(opts, "SpreadSort")
+      .value("No", jax_finufft_opts::SpreadSort::No)
+      .value("Yes", jax_finufft_opts::SpreadSort::Yes)
+      .value("Heuristic", jax_finufft_opts::SpreadSort::Heuristic)
+      .export_values();
+
+  py::enum_<jax_finufft_opts::SpreadThread>(opts, "SpreadThread")
+      .value("Auto", jax_finufft_opts::SpreadThread::Auto)
+      .value("Seq", jax_finufft_opts::SpreadThread::Seq)
+      .value("Parallel", jax_finufft_opts::SpreadThread::Parallel)
+      .export_values();
+
+  opts.def(
+      py::init<bool, bool, jax_finufft_opts::DebugLevel, jax_finufft_opts::DebugLevel, bool, int,
+               int, jax_finufft_opts::SpreadSort, bool, bool, double,
+               jax_finufft_opts::SpreadThread, int, int, int>(),
+      py::arg("modeord") = false, py::arg("chkbnds") = true,
+      py::arg("debug") = jax_finufft_opts::DebugLevel::Silent,
+      py::arg("spread_debug") = jax_finufft_opts::DebugLevel::Silent, py::arg("showwarn") = false,
+      py::arg("nthreads") = 0, py::arg("fftw") = int(FFTW_ESTIMATE),
+      py::arg("spread_sort") = jax_finufft_opts::SpreadSort::Heuristic,
+      py::arg("spread_kerevalmeth") = true, py::arg("spread_kerpad") = true,
+      py::arg("upsampfac") = 0.0, py::arg("spread_thread") = jax_finufft_opts::SpreadThread::Auto,
+      py::arg("maxbatchsize") = 0, py::arg("spread_nthr_atomic") = -1,
+      py::arg("spread_max_sp_size") = 0);
 }
 
 }  // namespace
diff --git a/lib/jax_finufft_cpu_opts.h b/lib/jax_finufft_cpu_opts.h
new file mode 100644
index 0000000..7078732
--- /dev/null
+++ b/lib/jax_finufft_cpu_opts.h
@@ -0,0 +1,72 @@
+#ifndef _JAX_FINUFFT_OPTS_H_
+#define _JAX_FINUFFT_OPTS_H_
+
+#include <fftw3.h>
+
+#include "finufft.h"
+#include "jax_finufft_cpu.h"
+
+namespace jax_finufft {
+
+struct jax_finufft_opts {
+  enum DebugLevel { Silent = 0, Verbose, Noisy };
+  enum FftwFlags {
+    Estimate = FFTW_ESTIMATE,
+    Measure = FFTW_MEASURE,
+    Patient = FFTW_PATIENT,
+    Exhaustive = FFTW_EXHAUSTIVE,
+    WisdomOnly = FFTW_WISDOM_ONLY
+  };
+  enum SpreadSort { No = 0, Yes, Heuristic };
+  enum SpreadThread { Auto = 0, Seq, Parallel };
+
+  finufft_opts opts;
+
+  jax_finufft_opts(
+      bool modeord,  // (type 1,2 only): 0 CMCL-style increasing mode order
+                     //                  1 FFT-style mode order
+      bool chkbnds,  // 0 don't check NU pts in [-3pi,3pi), 1 do (<few % slower)
+
+      // diagnostic opts...
+      DebugLevel debug,
+      DebugLevel spread_debug,  // spreader: 0 silent, 1 some timing/debug, or 2 tonnes
+      bool showwarn,            // 0 don't print warnings to stderr, 1 do
+
+      // algorithm performance opts...
+      int nthreads,                // number of threads to use, or 0 uses all available
+      int fftw,                    // plan flags to FFTW (FFTW_ESTIMATE=64, FFTW_MEASURE=0,...)
+      SpreadSort spread_sort,      // spreader: 0 don't sort, 1 do, or 2 heuristic choice
+      bool spread_kerevalmeth,     // spreader: 0 exp(sqrt()), 1 Horner piecewise poly (faster)
+      bool spread_kerpad,          // (exp(sqrt()) only): 0 don't pad kernel to 4n, 1 do
+      double upsampfac,            // upsampling ratio sigma: 2.0 std, 1.25 small FFT, 0.0 auto
+      SpreadThread spread_thread,  // (vectorized ntr>1 only): 0 auto, 1 seq multithreaded,
+                                   //                          2 parallel single-thread spread
+      int maxbatchsize,            // (vectorized ntr>1 only): max transform batch, 0 auto
+      int spread_nthr_atomic,      // if >=0, threads above which spreader OMP critical goes atomic
+      int spread_max_sp_size       // if >0, overrides spreader (dir=1) max subproblem size
+  ) {
+    default_opts<double>(&opts);
+
+    opts.modeord = modeord;
+    opts.chkbnds = chkbnds;
+
+    opts.debug = int(debug);
+    opts.spread_debug = int(spread_debug);
+    opts.showwarn = int(showwarn);
+
+    opts.nthreads = nthreads;
+    opts.fftw = fftw;
+    opts.spread_sort = spread_sort;
+    opts.spread_kerevalmeth = int(spread_kerevalmeth);
+    opts.spread_kerpad = int(spread_kerpad);
+    opts.upsampfac = upsampfac;
+    opts.spread_thread = int(spread_thread);
+    opts.maxbatchsize = maxbatchsize;
+    opts.spread_nthr_atomic = spread_nthr_atomic;
+    opts.spread_max_sp_size = spread_max_sp_size;
+  }
+};
+
+}  // namespace jax_finufft
+
+#endif