diff --git a/gopls/internal/golang/completion/completion.go b/gopls/internal/golang/completion/completion.go
index 6bf8ad8acde..e52dc9b73ec 100644
--- a/gopls/internal/golang/completion/completion.go
+++ b/gopls/internal/golang/completion/completion.go
@@ -38,6 +38,7 @@ import (
 	"golang.org/x/tools/gopls/internal/protocol"
 	"golang.org/x/tools/gopls/internal/settings"
 	goplsastutil "golang.org/x/tools/gopls/internal/util/astutil"
+	"golang.org/x/tools/gopls/internal/util/bug"
 	"golang.org/x/tools/gopls/internal/util/safetoken"
 	"golang.org/x/tools/gopls/internal/util/typesutil"
 	"golang.org/x/tools/internal/event"
@@ -134,6 +135,33 @@ func (i *CompletionItem) Snippet() string {
 	return i.InsertText
 }
 
+// addConversion wraps the existing completionItem in a conversion expression.
+// Only affects the receiver's InsertText and snippet fields, not the Label.
+// An empty conv argument has no effect.
+func (i *CompletionItem) addConversion(c *completer, conv conversionEdits) error {
+	if conv.prefix != "" {
+		// If we are in a selector, add an edit to place prefix before selector.
+		if sel := enclosingSelector(c.path, c.pos); sel != nil {
+			edits, err := c.editText(sel.Pos(), sel.Pos(), conv.prefix)
+			if err != nil {
+				return err
+			}
+			i.AdditionalTextEdits = append(i.AdditionalTextEdits, edits...)
+		} else {
+			// If there is no selector, just stick the prefix at the start.
+			i.InsertText = conv.prefix + i.InsertText
+			i.snippet.PrependText(conv.prefix)
+		}
+	}
+
+	if conv.suffix != "" {
+		i.InsertText += conv.suffix
+		i.snippet.WriteText(conv.suffix)
+	}
+
+	return nil
+}
+
 // Scoring constants are used for weighting the relevance of different candidates.
 const (
 	// stdScore is the base score for all completion items.
@@ -682,7 +710,7 @@ func (c *completer) collectCompletions(ctx context.Context) error {
 	}
 
 	// Struct literals are handled entirely separately.
-	if c.wantStructFieldCompletions() {
+	if wantStructFieldCompletions(c.enclosingCompositeLiteral) {
 		// If we are definitely completing a struct field name, deep completions
 		// don't make sense.
 		if c.enclosingCompositeLiteral.inKey {
@@ -1133,12 +1161,11 @@ func (c *completer) addFieldItems(fields *ast.FieldList) {
 	}
 }
 
-func (c *completer) wantStructFieldCompletions() bool {
-	clInfo := c.enclosingCompositeLiteral
-	if clInfo == nil {
+func wantStructFieldCompletions(enclosingCl *compLitInfo) bool {
+	if enclosingCl == nil {
 		return false
 	}
-	return is[*types.Struct](clInfo.clType) && (clInfo.inKey || clInfo.maybeInFieldName)
+	return is[*types.Struct](enclosingCl.clType) && (enclosingCl.inKey || enclosingCl.maybeInFieldName)
 }
 
 func (c *completer) wantTypeName() bool {
@@ -2039,8 +2066,7 @@ func enclosingFunction(path []ast.Node, info *types.Info) *funcInfo {
 	return nil
 }
 
-func (c *completer) expectedCompositeLiteralType() types.Type {
-	clInfo := c.enclosingCompositeLiteral
+func expectedCompositeLiteralType(clInfo *compLitInfo, pos token.Pos) types.Type {
 	switch t := clInfo.clType.(type) {
 	case *types.Slice:
 		if clInfo.inKey {
@@ -2079,7 +2105,7 @@ func (c *completer) expectedCompositeLiteralType() types.Type {
 
 			// The order of the literal fields must match the order in the struct definition.
 			// Find the element that the position belongs to and suggest that field's type.
-			if i := exprAtPos(c.pos, clInfo.cl.Elts); i < t.NumFields() {
+			if i := exprAtPos(pos, clInfo.cl.Elts); i < t.NumFields() {
 				return t.Field(i).Type()
 			}
 		}
@@ -2164,6 +2190,25 @@ type candidateInference struct {
 	// convertibleTo is a type our candidate type must be convertible to.
 	convertibleTo types.Type
 
+	// needsExactType is true if the candidate type must be exactly the type of
+	// the objType, e.g. an interface rather than it's implementors.
+	//
+	// This is necessary when objType is derived using reverse type inference:
+	// any different (but assignable) type may lead to different type inference,
+	// which may no longer be valid.
+	//
+	// For example, consider the following scenario:
+	//
+	//   func f[T any](x T) []T { return []T{x} }
+	//
+	//   var s []any = f(_)
+	//
+	// Reverse type inference would infer that the type at _ must be 'any', but
+	// that does not mean that any object in the lexical scope is valid: the type of
+	// the object must be *exactly* any, otherwise type inference will cause the
+	// slice assignment to fail.
+	needsExactType bool
+
 	// typeName holds information about the expected type name at
 	// position, if any.
 	typeName typeNameInference
@@ -2235,7 +2280,7 @@ func expectedCandidate(ctx context.Context, c *completer) (inf candidateInferenc
 	inf.typeName = expectTypeName(c)
 
 	if c.enclosingCompositeLiteral != nil {
-		inf.objType = c.expectedCompositeLiteralType()
+		inf.objType = expectedCompositeLiteralType(c.enclosingCompositeLiteral, c.pos)
 	}
 
 Nodes:
@@ -2259,34 +2304,21 @@ Nodes:
 				break Nodes
 			}
 		case *ast.AssignStmt:
-			// Only rank completions if you are on the right side of the token.
-			if c.pos > node.TokPos {
-				i := exprAtPos(c.pos, node.Rhs)
-				if i >= len(node.Lhs) {
-					i = len(node.Lhs) - 1
-				}
-				if tv, ok := c.pkg.TypesInfo().Types[node.Lhs[i]]; ok {
-					inf.objType = tv.Type
-				}
-
-				// If we have a single expression on the RHS, record the LHS
-				// assignees so we can favor multi-return function calls with
-				// matching result values.
-				if len(node.Rhs) <= 1 {
-					for _, lhs := range node.Lhs {
-						inf.assignees = append(inf.assignees, c.pkg.TypesInfo().TypeOf(lhs))
-					}
-				} else {
-					// Otherwise, record our single assignee, even if its type is
-					// not available. We use this info to downrank functions
-					// with the wrong number of result values.
-					inf.assignees = append(inf.assignees, c.pkg.TypesInfo().TypeOf(node.Lhs[i]))
-				}
-			}
+			objType, assignees := expectedAssignStmtTypes(c.pkg, node, c.pos)
+			inf.objType = objType
+			inf.assignees = assignees
 			return inf
 		case *ast.ValueSpec:
-			if node.Type != nil && c.pos > node.Type.End() {
-				inf.objType = c.pkg.TypesInfo().TypeOf(node.Type)
+			inf.objType = expectedValueSpecType(c.pkg, node, c.pos)
+			return
+		case *ast.ReturnStmt:
+			if c.enclosingFunc != nil {
+				inf.objType = expectedReturnStmtType(c.enclosingFunc.sig, node, c.pos)
+			}
+			return inf
+		case *ast.SendStmt:
+			if typ := expectedSendStmtType(c.pkg, node, c.pos); typ != nil {
+				inf.objType = typ
 			}
 			return inf
 		case *ast.CallExpr:
@@ -2299,22 +2331,27 @@ Nodes:
 					break Nodes
 				}
 
-				sig, _ := c.pkg.TypesInfo().Types[node.Fun].Type.(*types.Signature)
-
-				if sig != nil && sig.TypeParams().Len() > 0 {
-					// If we are completing a generic func call, re-check the call expression.
-					// This allows type param inference to work in cases like:
-					//
-					// func foo[T any](T) {}
-					// foo[int](<>) // <- get "int" completions instead of "T"
-					//
-					// TODO: remove this after https://go.dev/issue/52503
-					info := &types.Info{Types: make(map[ast.Expr]types.TypeAndValue)}
-					types.CheckExpr(c.pkg.FileSet(), c.pkg.Types(), node.Fun.Pos(), node.Fun, info)
-					sig, _ = info.Types[node.Fun].Type.(*types.Signature)
-				}
+				if sig, ok := c.pkg.TypesInfo().Types[node.Fun].Type.(*types.Signature); ok {
+					// Out of bounds arguments get no inference completion.
+					if !sig.Variadic() && exprAtPos(c.pos, node.Args) >= sig.Params().Len() {
+						return inf
+					}
+
+					if sig.TypeParams().Len() > 0 {
+						targs := c.getTypeArgs(node)
+						res := inferExpectedResultTypes(c, i)
+						substs := reverseInferTypeArgs(sig, targs, res)
+						inst := instantiate(sig, substs)
+						if inst != nil {
+							// TODO(jacobz): If partial signature instantiation becomes possible,
+							// make needsExactType only true if necessary.
+							// Currently, ambigious cases always resolve to a conversion expression
+							// wrapping the completion, which is occassionally superfluous.
+							inf.needsExactType = true
+							sig = inst
+						}
+					}
 
-				if sig != nil {
 					inf = c.expectedCallParamType(inf, node, sig)
 				}
 
@@ -2344,17 +2381,6 @@ Nodes:
 
 				return inf
 			}
-		case *ast.ReturnStmt:
-			if c.enclosingFunc != nil {
-				sig := c.enclosingFunc.sig
-				// Find signature result that corresponds to our return statement.
-				if resultIdx := exprAtPos(c.pos, node.Results); resultIdx < len(node.Results) {
-					if resultIdx < sig.Results().Len() {
-						inf.objType = sig.Results().At(resultIdx).Type()
-					}
-				}
-			}
-			return inf
 		case *ast.CaseClause:
 			if swtch, ok := findSwitchStmt(c.path[i+1:], c.pos, node).(*ast.SwitchStmt); ok {
 				if tv, ok := c.pkg.TypesInfo().Types[swtch.Tag]; ok {
@@ -2398,6 +2424,9 @@ Nodes:
 						inf.objType = ct
 						inf.typeName.wantTypeName = true
 						inf.typeName.isTypeParam = true
+						if typ := c.inferExpectedTypeArg(i+1, 0); typ != nil {
+							inf.objType = typ
+						}
 					}
 				}
 			}
@@ -2405,20 +2434,14 @@ Nodes:
 		case *ast.IndexListExpr:
 			if node.Lbrack < c.pos && c.pos <= node.Rbrack {
 				if tv, ok := c.pkg.TypesInfo().Types[node.X]; ok {
-					if ct := expectedConstraint(tv.Type, exprAtPos(c.pos, node.Indices)); ct != nil {
+					typeParamIdx := exprAtPos(c.pos, node.Indices)
+					if ct := expectedConstraint(tv.Type, typeParamIdx); ct != nil {
 						inf.objType = ct
 						inf.typeName.wantTypeName = true
 						inf.typeName.isTypeParam = true
-					}
-				}
-			}
-			return inf
-		case *ast.SendStmt:
-			// Make sure we are on right side of arrow (e.g. "foo <- <>").
-			if c.pos > node.Arrow+1 {
-				if tv, ok := c.pkg.TypesInfo().Types[node.Chan]; ok {
-					if ch, ok := tv.Type.Underlying().(*types.Chan); ok {
-						inf.objType = ch.Elem()
+						if typ := c.inferExpectedTypeArg(i+1, typeParamIdx); typ != nil {
+							inf.objType = typ
+						}
 					}
 				}
 			}
@@ -2457,6 +2480,267 @@ Nodes:
 	return inf
 }
 
+// inferExpectedResultTypes takes the index of a call expression within the completion
+// path and uses its surroundings to infer the expected result tuple of the call's signature.
+// Returns the signature result tuple as a slice, or nil if reverse type inference fails.
+//
+// # For example
+//
+// func generic[T any, U any](a T, b U) (T, U) { ... }
+//
+// var x TypeA
+// var y TypeB
+// x, y := generic(<cursor>, <cursor>)
+//
+// inferExpectedResultTypes can determine that the expected result type of the function is (TypeA, TypeB)
+func inferExpectedResultTypes(c *completer, callNodeIdx int) []types.Type {
+	callNode, ok := c.path[callNodeIdx].(*ast.CallExpr)
+	if !ok {
+		bug.Reportf("inferExpectedResultTypes given callNodeIndex: %v which is not a ast.CallExpr\n", callNodeIdx)
+		return nil
+	}
+
+	if len(c.path) <= callNodeIdx+1 {
+		return nil
+	}
+
+	var expectedResults []types.Type
+
+	// Check the parents of the call node to extract the expected result types of the call signature.
+	// Currently reverse inferences are only supported with the the following parent expressions,
+	// however this list isn't exhaustive.
+	switch node := c.path[callNodeIdx+1].(type) {
+	case *ast.KeyValueExpr:
+		enclosingCompositeLiteral := enclosingCompositeLiteral(c.path[callNodeIdx:], callNode.Pos(), c.pkg.TypesInfo())
+		if !wantStructFieldCompletions(enclosingCompositeLiteral) {
+			expectedResults = append(expectedResults, expectedCompositeLiteralType(enclosingCompositeLiteral, callNode.Pos()))
+		}
+	case *ast.AssignStmt:
+		objType, assignees := expectedAssignStmtTypes(c.pkg, node, c.pos)
+		if len(assignees) > 0 {
+			return assignees
+		} else if objType != nil {
+			expectedResults = append(expectedResults, objType)
+		}
+	case *ast.ValueSpec:
+		if resultType := expectedValueSpecType(c.pkg, node, c.pos); resultType != nil {
+			expectedResults = append(expectedResults, resultType)
+		}
+	case *ast.SendStmt:
+		if resultType := expectedSendStmtType(c.pkg, node, c.pos); resultType != nil {
+			expectedResults = append(expectedResults, resultType)
+		}
+	case *ast.ReturnStmt:
+		if c.enclosingFunc == nil {
+			return nil
+		}
+
+		// As a special case for reverse call inference in
+		//
+		// return foo(<cursor>)
+		//
+		// Pull the result type from the enclosing function
+		if exprAtPos(c.pos, node.Results) == 0 {
+			if callSig := c.pkg.TypesInfo().Types[callNode.Fun].Type.(*types.Signature); callSig != nil {
+				enclosingResults := c.enclosingFunc.sig.Results()
+				if callSig.Results().Len() == enclosingResults.Len() {
+					expectedResults = make([]types.Type, enclosingResults.Len())
+					for i := range enclosingResults.Len() {
+						expectedResults[i] = enclosingResults.At(i).Type()
+					}
+					return expectedResults
+				}
+			}
+		}
+
+		if resultType := expectedReturnStmtType(c.enclosingFunc.sig, node, c.pos); resultType != nil {
+			expectedResults = append(expectedResults, resultType)
+		}
+	case *ast.CallExpr:
+		// TODO(jacobz): This is a difficult case because the normal CallExpr candidateInference
+		// leans on control flow which is inaccessible in this helper function.
+		// It would probably take a significant refactor to a recursive solution to make this case
+		// work cleanly. For now it's unimplemented.
+	}
+	return expectedResults
+}
+
+// expectedSendStmtType return the expected type at the position.
+// Returns nil if unknown.
+func expectedSendStmtType(pkg *cache.Package, node *ast.SendStmt, pos token.Pos) types.Type {
+	// Make sure we are on right side of arrow (e.g. "foo <- <>").
+	if pos > node.Arrow+1 {
+		if tv, ok := pkg.TypesInfo().Types[node.Chan]; ok {
+			if ch, ok := tv.Type.Underlying().(*types.Chan); ok {
+				return ch.Elem()
+			}
+		}
+	}
+	return nil
+}
+
+// expectedValueSpecType returns the expected type of a ValueSpec at the query
+// position.
+func expectedValueSpecType(pkg *cache.Package, node *ast.ValueSpec, pos token.Pos) types.Type {
+	if node.Type != nil && pos > node.Type.End() {
+		return pkg.TypesInfo().TypeOf(node.Type)
+	}
+	return nil
+}
+
+// expectedAssignStmtTypes analyzes the provided assignStmt, and checks
+// to see if the provided pos is within a RHS expresison. If so, it report
+// the expected type of that expression, and the LHS type(s) to which it
+// is being assigned.
+func expectedAssignStmtTypes(pkg *cache.Package, node *ast.AssignStmt, pos token.Pos) (objType types.Type, assignees []types.Type) {
+	// Only rank completions if you are on the right side of the token.
+	if pos > node.TokPos {
+		i := exprAtPos(pos, node.Rhs)
+		if i >= len(node.Lhs) {
+			i = len(node.Lhs) - 1
+		}
+		if tv, ok := pkg.TypesInfo().Types[node.Lhs[i]]; ok {
+			objType = tv.Type
+		}
+
+		// If we have a single expression on the RHS, record the LHS
+		// assignees so we can favor multi-return function calls with
+		// matching result values.
+		if len(node.Rhs) <= 1 {
+			for _, lhs := range node.Lhs {
+				assignees = append(assignees, pkg.TypesInfo().TypeOf(lhs))
+			}
+		} else {
+			// Otherwise, record our single assignee, even if its type is
+			// not available. We use this info to downrank functions
+			// with the wrong number of result values.
+			assignees = append(assignees, pkg.TypesInfo().TypeOf(node.Lhs[i]))
+		}
+	}
+	return objType, assignees
+}
+
+// expectedReturnStmtType returns the expected type of a return statement.
+// Returns nil if enclosingSig is nil.
+func expectedReturnStmtType(enclosingSig *types.Signature, node *ast.ReturnStmt, pos token.Pos) types.Type {
+	if enclosingSig != nil {
+		if resultIdx := exprAtPos(pos, node.Results); resultIdx < len(node.Results) {
+			return enclosingSig.Results().At(resultIdx).Type()
+		}
+	}
+	return nil
+}
+
+// Returns the number of type arguments in a callExpr
+func (c *completer) getTypeArgs(callExpr *ast.CallExpr) []types.Type {
+	var targs []types.Type
+	switch fun := callExpr.Fun.(type) {
+	case *ast.IndexListExpr:
+		for i := range fun.Indices {
+			if typ, ok := c.pkg.TypesInfo().Types[fun.Indices[i]]; ok && typeIsValid(typ.Type) {
+				targs = append(targs, typ.Type)
+			}
+		}
+	case *ast.IndexExpr:
+		if typ, ok := c.pkg.TypesInfo().Types[fun.Index]; ok && typeIsValid(typ.Type) {
+			targs = []types.Type{typ.Type}
+		}
+	}
+	return targs
+}
+
+// reverseInferTypeArgs takes a generic signature, a list of passed type arguments, and the expected concrete return types
+// inferred from the signature's call site. If possible, it returns a list of types that could be used as the type arguments
+// to the signature. If not possible, it returns nil.
+//
+// Does not panic if any of the arguments are nil.
+func reverseInferTypeArgs(sig *types.Signature, typeArgs []types.Type, expectedResults []types.Type) []types.Type {
+	if len(expectedResults) == 0 || sig == nil || sig.TypeParams().Len() == 0 || sig.Results().Len() != len(expectedResults) {
+		return nil
+	}
+
+	tparams := make([]*types.TypeParam, sig.TypeParams().Len())
+	for i := range sig.TypeParams().Len() {
+		tparams[i] = sig.TypeParams().At(i)
+	}
+
+	for i := len(typeArgs); i < sig.TypeParams().Len(); i++ {
+		typeArgs = append(typeArgs, nil)
+	}
+
+	u := newUnifier(tparams, typeArgs)
+	for i, assignee := range expectedResults {
+		// Unify does not check the constraints of the type parameters.
+		// Checks must be applied after.
+		if !u.unify(sig.Results().At(i).Type(), assignee, unifyModeExact) {
+			return nil
+		}
+	}
+
+	substs := make([]types.Type, sig.TypeParams().Len())
+	for i := 0; i < sig.TypeParams().Len(); i++ {
+		if sub := u.handles[sig.TypeParams().At(i)]; sub != nil && *sub != nil {
+			// Ensure the inferred subst is assignable to the type parameter's constraint.
+			if !assignableTo(*sub, sig.TypeParams().At(i).Constraint()) {
+				return nil
+			}
+			substs[i] = *sub
+		}
+	}
+	return substs
+}
+
+// inferExpectedTypeArg gives a type param candidateInference based on the surroundings of it's call site.
+// If successful, the inf parameter is returned with only it's objType field updated.
+//
+// callNodeIdx is the index within the completion path of the type parameter's parent call expression.
+// typeParamIdx is the index of the type parameter at the completion pos.
+func (c *completer) inferExpectedTypeArg(callNodeIdx int, typeParamIdx int) types.Type {
+	if len(c.path) <= callNodeIdx {
+		return nil
+	}
+
+	callNode, ok := c.path[callNodeIdx].(*ast.CallExpr)
+	if !ok {
+		return nil
+	}
+
+	// Infer the type parameters in a function call based on it's context
+	sig := c.pkg.TypesInfo().Types[callNode.Fun].Type.(*types.Signature)
+	expectedResults := inferExpectedResultTypes(c, callNodeIdx)
+	if typeParamIdx < 0 || typeParamIdx >= sig.TypeParams().Len() {
+		return nil
+	}
+	substs := reverseInferTypeArgs(sig, nil, expectedResults)
+	if substs == nil || substs[typeParamIdx] == nil {
+		return nil
+	}
+
+	return substs[typeParamIdx]
+}
+
+// Instantiates a signature with a set of type parameters.
+// Wrapper around types.Instantiate but bad arguments won't cause a panic.
+func instantiate(sig *types.Signature, substs []types.Type) *types.Signature {
+	if substs == nil || sig == nil || len(substs) != sig.TypeParams().Len() {
+		return nil
+	}
+
+	for i := range substs {
+		if substs[i] == nil {
+			substs[i] = sig.TypeParams().At(i)
+		}
+	}
+
+	if inst, err := types.Instantiate(nil, sig, substs, true); err == nil {
+		if inst, ok := inst.(*types.Signature); ok {
+			return inst
+		}
+	}
+
+	return nil
+}
+
 func (c *completer) expectedCallParamType(inf candidateInference, node *ast.CallExpr, sig *types.Signature) candidateInference {
 	numParams := sig.Params().Len()
 	if numParams == 0 {
@@ -2875,7 +3159,7 @@ func (c *completer) matchingCandidate(cand *candidate) bool {
 	}
 
 	// Bail out early if we are completing a field name in a composite literal.
-	if v, ok := cand.obj.(*types.Var); ok && v.IsField() && c.wantStructFieldCompletions() {
+	if v, ok := cand.obj.(*types.Var); ok && v.IsField() && wantStructFieldCompletions(c.enclosingCompositeLiteral) {
 		return true
 	}
 
@@ -2972,6 +3256,14 @@ func (ci *candidateInference) candTypeMatches(cand *candidate) bool {
 				cand.mods = append(cand.mods, takeDotDotDot)
 			}
 
+			// Candidate matches, but isn't exactly identical to the expected type.
+			// Apply a conversion to allow it to match.
+			if ci.needsExactType && !types.Identical(candType, expType) {
+				cand.convertTo = expType
+				// Ranks barely lower if it needs a conversion, even though it's perfectly valid.
+				cand.score *= 0.95
+			}
+
 			// Lower candidate score for untyped conversions. This avoids
 			// ranking untyped constants above candidates with an exact type
 			// match. Don't lower score of builtin constants, e.g. "true".
@@ -3161,6 +3453,9 @@ func (c *completer) matchingTypeName(cand *candidate) bool {
 		return false
 	}
 
+	wantExactTypeParam := c.inference.typeName.isTypeParam &&
+		c.inference.typeName.wantTypeName && c.inference.needsExactType
+
 	typeMatches := func(candType types.Type) bool {
 		// Take into account any type name modifier prefixes.
 		candType = c.inference.applyTypeNameModifiers(candType)
@@ -3179,6 +3474,13 @@ func (c *completer) matchingTypeName(cand *candidate) bool {
 			}
 		}
 
+		// Suggest the exact type when performing reverse type inference.
+		// x = Foo[<>]()
+		// Where x is an interface kind, only suggest the interface type rather than its implementors
+		if wantExactTypeParam && types.Identical(candType, c.inference.objType) {
+			return true
+		}
+
 		if c.inference.typeName.wantComparable && !types.Comparable(candType) {
 			return false
 		}
diff --git a/gopls/internal/golang/completion/format.go b/gopls/internal/golang/completion/format.go
index baf0890497b..872025949fb 100644
--- a/gopls/internal/golang/completion/format.go
+++ b/gopls/internal/golang/completion/format.go
@@ -196,24 +196,9 @@ Suffixes:
 	}
 
 	if cand.convertTo != nil {
-		typeName := types.TypeString(cand.convertTo, c.qf)
-
-		switch t := cand.convertTo.(type) {
-		// We need extra parens when casting to these types. For example,
-		// we need "(*int)(foo)", not "*int(foo)".
-		case *types.Pointer, *types.Signature:
-			typeName = "(" + typeName + ")"
-		case *types.Basic:
-			// If the types are incompatible (as determined by typeMatches), then we
-			// must need a conversion here. However, if the target type is untyped,
-			// don't suggest converting to e.g. "untyped float" (golang/go#62141).
-			if t.Info()&types.IsUntyped != 0 {
-				typeName = types.TypeString(types.Default(cand.convertTo), c.qf)
-			}
-		}
-
-		prefix = typeName + "(" + prefix
-		suffix = ")"
+		conv := c.formatConversion(cand.convertTo)
+		prefix = conv.prefix + prefix
+		suffix = conv.suffix
 	}
 
 	if prefix != "" {
@@ -288,6 +273,38 @@ Suffixes:
 	return item, nil
 }
 
+// conversionEdits represents the string edits needed to make a type conversion
+// of an expression.
+type conversionEdits struct {
+	prefix, suffix string
+}
+
+// formatConversion returns the edits needed to make a type conversion
+// expression, including parentheses if necessary.
+//
+// Returns empty conversionEdits if convertTo is nil.
+func (c *completer) formatConversion(convertTo types.Type) conversionEdits {
+	if convertTo == nil {
+		return conversionEdits{}
+	}
+
+	typeName := types.TypeString(convertTo, c.qf)
+	switch t := convertTo.(type) {
+	// We need extra parens when casting to these types. For example,
+	// we need "(*int)(foo)", not "*int(foo)".
+	case *types.Pointer, *types.Signature:
+		typeName = "(" + typeName + ")"
+	case *types.Basic:
+		// If the types are incompatible (as determined by typeMatches), then we
+		// must need a conversion here. However, if the target type is untyped,
+		// don't suggest converting to e.g. "untyped float" (golang/go#62141).
+		if t.Info()&types.IsUntyped != 0 {
+			typeName = types.TypeString(types.Default(convertTo), c.qf)
+		}
+	}
+	return conversionEdits{prefix: typeName + "(", suffix: ")"}
+}
+
 // importEdits produces the text edits necessary to add the given import to the current file.
 func (c *completer) importEdits(imp *importInfo) ([]protocol.TextEdit, error) {
 	if imp == nil {
diff --git a/gopls/internal/golang/completion/literal.go b/gopls/internal/golang/completion/literal.go
index 82f123048e8..129045d99e2 100644
--- a/gopls/internal/golang/completion/literal.go
+++ b/gopls/internal/golang/completion/literal.go
@@ -73,15 +73,21 @@ func (c *completer) literal(ctx context.Context, literalType types.Type, imp *im
 		cand.addressable = true
 	}
 
-	if !c.matchingCandidate(&cand) || cand.convertTo != nil {
+	// Only suggest a literal conversion if the exact type is known.
+	if !c.matchingCandidate(&cand) || (cand.convertTo != nil && !c.inference.needsExactType) {
 		return
 	}
 
 	var (
-		qf  = c.qf
-		sel = enclosingSelector(c.path, c.pos)
+		qf         = c.qf
+		sel        = enclosingSelector(c.path, c.pos)
+		conversion conversionEdits
 	)
 
+	if cand.convertTo != nil {
+		conversion = c.formatConversion(cand.convertTo)
+	}
+
 	// Don't qualify the type name if we are in a selector expression
 	// since the package name is already present.
 	if sel != nil {
@@ -129,13 +135,18 @@ func (c *completer) literal(ctx context.Context, literalType types.Type, imp *im
 
 		switch t := literalType.Underlying().(type) {
 		case *types.Struct, *types.Array, *types.Slice, *types.Map:
-			c.compositeLiteral(t, snip.Clone(), typeName, float64(score), addlEdits)
+			item := c.compositeLiteral(t, snip.Clone(), typeName, float64(score), addlEdits)
+			item.addConversion(c, conversion)
+			c.items = append(c.items, item)
 		case *types.Signature:
 			// Add a literal completion for a signature type that implements
 			// an interface. For example, offer "http.HandlerFunc()" when
 			// expected type is "http.Handler".
 			if expType != nil && types.IsInterface(expType) {
-				c.basicLiteral(t, snip.Clone(), typeName, float64(score), addlEdits)
+				if item, ok := c.basicLiteral(t, snip.Clone(), typeName, float64(score), addlEdits); ok {
+					item.addConversion(c, conversion)
+					c.items = append(c.items, item)
+				}
 			}
 		case *types.Basic:
 			// Add a literal completion for basic types that implement our
@@ -143,7 +154,10 @@ func (c *completer) literal(ctx context.Context, literalType types.Type, imp *im
 			// implements http.FileSystem), or are identical to our expected
 			// type (i.e. yielding a type conversion such as "float64()").
 			if expType != nil && (types.IsInterface(expType) || types.Identical(expType, literalType)) {
-				c.basicLiteral(t, snip.Clone(), typeName, float64(score), addlEdits)
+				if item, ok := c.basicLiteral(t, snip.Clone(), typeName, float64(score), addlEdits); ok {
+					item.addConversion(c, conversion)
+					c.items = append(c.items, item)
+				}
 			}
 		}
 	}
@@ -155,11 +169,15 @@ func (c *completer) literal(ctx context.Context, literalType types.Type, imp *im
 		switch literalType.Underlying().(type) {
 		case *types.Slice:
 			// The second argument to "make()" for slices is required, so default to "0".
-			c.makeCall(snip.Clone(), typeName, "0", float64(score), addlEdits)
+			item := c.makeCall(snip.Clone(), typeName, "0", float64(score), addlEdits)
+			item.addConversion(c, conversion)
+			c.items = append(c.items, item)
 		case *types.Map, *types.Chan:
 			// Maps and channels don't require the second argument, so omit
 			// to keep things simple for now.
-			c.makeCall(snip.Clone(), typeName, "", float64(score), addlEdits)
+			item := c.makeCall(snip.Clone(), typeName, "", float64(score), addlEdits)
+			item.addConversion(c, conversion)
+			c.items = append(c.items, item)
 		}
 	}
 
@@ -167,7 +185,10 @@ func (c *completer) literal(ctx context.Context, literalType types.Type, imp *im
 	if score := c.matcher.Score("func"); !cand.hasMod(reference) && score > 0 && (expType == nil || !types.IsInterface(expType)) {
 		switch t := literalType.Underlying().(type) {
 		case *types.Signature:
-			c.functionLiteral(ctx, t, float64(score))
+			if item, ok := c.functionLiteral(ctx, t, float64(score)); ok {
+				item.addConversion(c, conversion)
+				c.items = append(c.items, item)
+			}
 		}
 	}
 }
@@ -178,9 +199,9 @@ func (c *completer) literal(ctx context.Context, literalType types.Type, imp *im
 // correct type, so scale down highScore.
 const literalCandidateScore = highScore / 2
 
-// functionLiteral adds a function literal completion item for the
-// given signature.
-func (c *completer) functionLiteral(ctx context.Context, sig *types.Signature, matchScore float64) {
+// functionLiteral returns a function literal completion item for the
+// given signature, if applicable.
+func (c *completer) functionLiteral(ctx context.Context, sig *types.Signature, matchScore float64) (CompletionItem, bool) {
 	snip := &snippet.Builder{}
 	snip.WriteText("func(")
 
@@ -216,7 +237,7 @@ func (c *completer) functionLiteral(ctx context.Context, sig *types.Signature, m
 				if ctx.Err() == nil {
 					event.Error(ctx, "formatting var type", err)
 				}
-				return
+				return CompletionItem{}, false
 			}
 			name = abbreviateTypeName(typeName)
 		}
@@ -284,7 +305,7 @@ func (c *completer) functionLiteral(ctx context.Context, sig *types.Signature, m
 				if ctx.Err() == nil {
 					event.Error(ctx, "formatting var type", err)
 				}
-				return
+				return CompletionItem{}, false
 			}
 			if sig.Variadic() && i == sig.Params().Len()-1 {
 				typeStr = strings.Replace(typeStr, "[]", "...", 1)
@@ -342,7 +363,7 @@ func (c *completer) functionLiteral(ctx context.Context, sig *types.Signature, m
 			if ctx.Err() == nil {
 				event.Error(ctx, "formatting var type", err)
 			}
-			return
+			return CompletionItem{}, false
 		}
 		if tp, ok := types.Unalias(r.Type()).(*types.TypeParam); ok && !c.typeParamInScope(tp) {
 			snip.WritePlaceholder(func(snip *snippet.Builder) {
@@ -360,12 +381,12 @@ func (c *completer) functionLiteral(ctx context.Context, sig *types.Signature, m
 	snip.WriteFinalTabstop()
 	snip.WriteText("}")
 
-	c.items = append(c.items, CompletionItem{
+	return CompletionItem{
 		Label:   "func(...) {}",
 		Score:   matchScore * literalCandidateScore,
 		Kind:    protocol.VariableCompletion,
 		snippet: snip,
-	})
+	}, true
 }
 
 // conventionalAcronyms contains conventional acronyms for type names
@@ -407,9 +428,9 @@ func abbreviateTypeName(s string) string {
 	return golang.AbbreviateVarName(s)
 }
 
-// compositeLiteral adds a composite literal completion item for the given typeName.
+// compositeLiteral returns a composite literal completion item for the given typeName.
 // T is an (unnamed, unaliased) struct, array, slice, or map type.
-func (c *completer) compositeLiteral(T types.Type, snip *snippet.Builder, typeName string, matchScore float64, edits []protocol.TextEdit) {
+func (c *completer) compositeLiteral(T types.Type, snip *snippet.Builder, typeName string, matchScore float64, edits []protocol.TextEdit) CompletionItem {
 	snip.WriteText("{")
 	// Don't put the tab stop inside the composite literal curlies "{}"
 	// for structs that have no accessible fields.
@@ -420,22 +441,24 @@ func (c *completer) compositeLiteral(T types.Type, snip *snippet.Builder, typeNa
 
 	nonSnippet := typeName + "{}"
 
-	c.items = append(c.items, CompletionItem{
+	return CompletionItem{
 		Label:               nonSnippet,
 		InsertText:          nonSnippet,
 		Score:               matchScore * literalCandidateScore,
 		Kind:                protocol.VariableCompletion,
 		AdditionalTextEdits: edits,
 		snippet:             snip,
-	})
+	}
 }
 
-// basicLiteral adds a literal completion item for the given basic
+// basicLiteral returns a literal completion item for the given basic
 // type name typeName.
-func (c *completer) basicLiteral(T types.Type, snip *snippet.Builder, typeName string, matchScore float64, edits []protocol.TextEdit) {
+//
+// If T is untyped, this function returns false.
+func (c *completer) basicLiteral(T types.Type, snip *snippet.Builder, typeName string, matchScore float64, edits []protocol.TextEdit) (CompletionItem, bool) {
 	// Never give type conversions like "untyped int()".
 	if isUntyped(T) {
-		return
+		return CompletionItem{}, false
 	}
 
 	snip.WriteText("(")
@@ -444,7 +467,7 @@ func (c *completer) basicLiteral(T types.Type, snip *snippet.Builder, typeName s
 
 	nonSnippet := typeName + "()"
 
-	c.items = append(c.items, CompletionItem{
+	return CompletionItem{
 		Label:               nonSnippet,
 		InsertText:          nonSnippet,
 		Detail:              T.String(),
@@ -452,11 +475,11 @@ func (c *completer) basicLiteral(T types.Type, snip *snippet.Builder, typeName s
 		Kind:                protocol.VariableCompletion,
 		AdditionalTextEdits: edits,
 		snippet:             snip,
-	})
+	}, true
 }
 
-// makeCall adds a completion item for a "make()" call given a specific type.
-func (c *completer) makeCall(snip *snippet.Builder, typeName string, secondArg string, matchScore float64, edits []protocol.TextEdit) {
+// makeCall returns a completion item for a "make()" call given a specific type.
+func (c *completer) makeCall(snip *snippet.Builder, typeName string, secondArg string, matchScore float64, edits []protocol.TextEdit) CompletionItem {
 	// Keep it simple and don't add any placeholders for optional "make()" arguments.
 
 	snip.PrependText("make(")
@@ -478,14 +501,15 @@ func (c *completer) makeCall(snip *snippet.Builder, typeName string, secondArg s
 	}
 	nonSnippet.WriteByte(')')
 
-	c.items = append(c.items, CompletionItem{
-		Label:               nonSnippet.String(),
-		InsertText:          nonSnippet.String(),
-		Score:               matchScore * literalCandidateScore,
+	return CompletionItem{
+		Label:      nonSnippet.String(),
+		InsertText: nonSnippet.String(),
+		// make() should be just below other literal completions
+		Score:               matchScore * literalCandidateScore * 0.99,
 		Kind:                protocol.FunctionCompletion,
 		AdditionalTextEdits: edits,
 		snippet:             snip,
-	})
+	}
 }
 
 // Create a snippet for a type name where type params become placeholders.
diff --git a/gopls/internal/golang/completion/snippet.go b/gopls/internal/golang/completion/snippet.go
index 8df81f87672..fe346203120 100644
--- a/gopls/internal/golang/completion/snippet.go
+++ b/gopls/internal/golang/completion/snippet.go
@@ -13,7 +13,7 @@ import (
 
 // structFieldSnippet calculates the snippet for struct literal field names.
 func (c *completer) structFieldSnippet(cand candidate, detail string, snip *snippet.Builder) {
-	if !c.wantStructFieldCompletions() {
+	if !wantStructFieldCompletions(c.enclosingCompositeLiteral) {
 		return
 	}
 
diff --git a/gopls/internal/golang/completion/unify.go b/gopls/internal/golang/completion/unify.go
new file mode 100644
index 00000000000..8f4a1d3cbe0
--- /dev/null
+++ b/gopls/internal/golang/completion/unify.go
@@ -0,0 +1,710 @@
+// Below was copied from go/types/unify.go on September 24, 2024,
+// and combined with snippets from other files as well.
+// It is copied to implement unification for code completion inferences,
+// in lieu of an official type unification API.
+//
+// TODO: When such an API is available, the code below should deleted.
+//
+// Due to complexity of extracting private types from the go/types package,
+// the unifier does not fully implement interface unification.
+//
+// The code has been modified to compile without introducing any key functionality changes.
+//
+
+// Copyright 2020 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// This file implements type unification.
+//
+// Type unification attempts to make two types x and y structurally
+// equivalent by determining the types for a given list of (bound)
+// type parameters which may occur within x and y. If x and y are
+// structurally different (say []T vs chan T), or conflicting
+// types are determined for type parameters, unification fails.
+// If unification succeeds, as a side-effect, the types of the
+// bound type parameters may be determined.
+//
+// Unification typically requires multiple calls u.unify(x, y) to
+// a given unifier u, with various combinations of types x and y.
+// In each call, additional type parameter types may be determined
+// as a side effect and recorded in u.
+// If a call fails (returns false), unification fails.
+//
+// In the unification context, structural equivalence of two types
+// ignores the difference between a defined type and its underlying
+// type if one type is a defined type and the other one is not.
+// It also ignores the difference between an (external, unbound)
+// type parameter and its core type.
+// If two types are not structurally equivalent, they cannot be Go
+// identical types. On the other hand, if they are structurally
+// equivalent, they may be Go identical or at least assignable, or
+// they may be in the type set of a constraint.
+// Whether they indeed are identical or assignable is determined
+// upon instantiation and function argument passing.
+
+package completion
+
+import (
+	"fmt"
+	"go/types"
+	"strings"
+)
+
+const (
+	// Upper limit for recursion depth. Used to catch infinite recursions
+	// due to implementation issues (e.g., see issues go.dev/issue/48619, go.dev/issue/48656).
+	unificationDepthLimit = 50
+
+	// Whether to panic when unificationDepthLimit is reached.
+	// If disabled, a recursion depth overflow results in a (quiet)
+	// unification failure.
+	panicAtUnificationDepthLimit = true
+
+	// If enableCoreTypeUnification is set, unification will consider
+	// the core types, if any, of non-local (unbound) type parameters.
+	enableCoreTypeUnification = true
+)
+
+// A unifier maintains a list of type parameters and
+// corresponding types inferred for each type parameter.
+// A unifier is created by calling newUnifier.
+type unifier struct {
+	// handles maps each type parameter to its inferred type through
+	// an indirection *Type called (inferred type) "handle".
+	// Initially, each type parameter has its own, separate handle,
+	// with a nil (i.e., not yet inferred) type.
+	// After a type parameter P is unified with a type parameter Q,
+	// P and Q share the same handle (and thus type). This ensures
+	// that inferring the type for a given type parameter P will
+	// automatically infer the same type for all other parameters
+	// unified (joined) with P.
+	handles map[*types.TypeParam]*types.Type
+	depth   int // recursion depth during unification
+}
+
+// newUnifier returns a new unifier initialized with the given type parameter
+// and corresponding type argument lists. The type argument list may be shorter
+// than the type parameter list, and it may contain nil types. Matching type
+// parameters and arguments must have the same index.
+func newUnifier(tparams []*types.TypeParam, targs []types.Type) *unifier {
+	handles := make(map[*types.TypeParam]*types.Type, len(tparams))
+	// Allocate all handles up-front: in a correct program, all type parameters
+	// must be resolved and thus eventually will get a handle.
+	// Also, sharing of handles caused by unified type parameters is rare and
+	// so it's ok to not optimize for that case (and delay handle allocation).
+	for i, x := range tparams {
+		var t types.Type
+		if i < len(targs) {
+			t = targs[i]
+		}
+		handles[x] = &t
+	}
+	return &unifier{handles, 0}
+}
+
+// unifyMode controls the behavior of the unifier.
+type unifyMode uint
+
+const (
+	// If unifyModeAssign is set, we are unifying types involved in an assignment:
+	// they may match inexactly at the top, but element types must match
+	// exactly.
+	unifyModeAssign unifyMode = 1 << iota
+
+	// If unifyModeExact is set, types unify if they are identical (or can be
+	// made identical with suitable arguments for type parameters).
+	// Otherwise, a named type and a type literal unify if their
+	// underlying types unify, channel directions are ignored, and
+	// if there is an interface, the other type must implement the
+	// interface.
+	unifyModeExact
+)
+
+// This function was copied from go/types/unify.go
+//
+// unify attempts to unify x and y and reports whether it succeeded.
+// As a side-effect, types may be inferred for type parameters.
+// The mode parameter controls how types are compared.
+func (u *unifier) unify(x, y types.Type, mode unifyMode) bool {
+	return u.nify(x, y, mode)
+}
+
+type typeParamsById []*types.TypeParam
+
+// join unifies the given type parameters x and y.
+// If both type parameters already have a type associated with them
+// and they are not joined, join fails and returns false.
+func (u *unifier) join(x, y *types.TypeParam) bool {
+	switch hx, hy := u.handles[x], u.handles[y]; {
+	case hx == hy:
+		// Both type parameters already share the same handle. Nothing to do.
+	case *hx != nil && *hy != nil:
+		// Both type parameters have (possibly different) inferred types. Cannot join.
+		return false
+	case *hx != nil:
+		// Only type parameter x has an inferred type. Use handle of x.
+		u.setHandle(y, hx)
+	// This case is treated like the default case.
+	// case *hy != nil:
+	// 	// Only type parameter y has an inferred type. Use handle of y.
+	//	u.setHandle(x, hy)
+	default:
+		// Neither type parameter has an inferred type. Use handle of y.
+		u.setHandle(x, hy)
+	}
+	return true
+}
+
+// asBoundTypeParam returns x.(*types.TypeParam) if x is a type parameter recorded with u.
+// Otherwise, the result is nil.
+func (u *unifier) asBoundTypeParam(x types.Type) *types.TypeParam {
+	if x, _ := types.Unalias(x).(*types.TypeParam); x != nil {
+		if _, found := u.handles[x]; found {
+			return x
+		}
+	}
+	return nil
+}
+
+// setHandle sets the handle for type parameter x
+// (and all its joined type parameters) to h.
+func (u *unifier) setHandle(x *types.TypeParam, h *types.Type) {
+	hx := u.handles[x]
+	for y, hy := range u.handles {
+		if hy == hx {
+			u.handles[y] = h
+		}
+	}
+}
+
+// at returns the (possibly nil) type for type parameter x.
+func (u *unifier) at(x *types.TypeParam) types.Type {
+	return *u.handles[x]
+}
+
+// set sets the type t for type parameter x;
+// t must not be nil.
+func (u *unifier) set(x *types.TypeParam, t types.Type) {
+	*u.handles[x] = t
+}
+
+// unknowns returns the number of type parameters for which no type has been set yet.
+func (u *unifier) unknowns() int {
+	n := 0
+	for _, h := range u.handles {
+		if *h == nil {
+			n++
+		}
+	}
+	return n
+}
+
+// inferred returns the list of inferred types for the given type parameter list.
+// The result is never nil and has the same length as tparams; result types that
+// could not be inferred are nil. Corresponding type parameters and result types
+// have identical indices.
+func (u *unifier) inferred(tparams []*types.TypeParam) []types.Type {
+	list := make([]types.Type, len(tparams))
+	for i, x := range tparams {
+		list[i] = u.at(x)
+	}
+	return list
+}
+
+// asInterface returns the underlying type of x as an interface if
+// it is a non-type parameter interface. Otherwise it returns nil.
+func asInterface(x types.Type) (i *types.Interface) {
+	if _, ok := types.Unalias(x).(*types.TypeParam); !ok {
+		i, _ = x.Underlying().(*types.Interface)
+	}
+	return i
+}
+
+func isTypeParam(t types.Type) bool {
+	_, ok := types.Unalias(t).(*types.TypeParam)
+	return ok
+}
+
+func asNamed(t types.Type) *types.Named {
+	n, _ := types.Unalias(t).(*types.Named)
+	return n
+}
+
+func isTypeLit(t types.Type) bool {
+	switch types.Unalias(t).(type) {
+	case *types.Named, *types.TypeParam:
+		return false
+	}
+	return true
+}
+
+// identicalOrigin reports whether x and y originated in the same declaration.
+func identicalOrigin(x, y *types.Named) bool {
+	// TODO(gri) is this correct?
+	return x.Origin().Obj() == y.Origin().Obj()
+}
+
+func match(x, y types.Type) types.Type {
+	// Common case: we don't have channels.
+	if types.Identical(x, y) {
+		return x
+	}
+
+	// We may have channels that differ in direction only.
+	if x, _ := x.(*types.Chan); x != nil {
+		if y, _ := y.(*types.Chan); y != nil && types.Identical(x.Elem(), y.Elem()) {
+			// We have channels that differ in direction only.
+			// If there's an unrestricted channel, select the restricted one.
+			switch {
+			case x.Dir() == types.SendRecv:
+				return y
+			case y.Dir() == types.SendRecv:
+				return x
+			}
+		}
+	}
+
+	// types are different
+	return nil
+}
+
+func coreType(t types.Type) types.Type {
+	t = types.Unalias(t)
+	tpar, _ := t.(*types.TypeParam)
+	if tpar == nil {
+		return t.Underlying()
+	}
+
+	return nil
+}
+
+func sameId(obj *types.Var, pkg *types.Package, name string, foldCase bool) bool {
+	// If we don't care about capitalization, we also ignore packages.
+	if foldCase && strings.EqualFold(obj.Name(), name) {
+		return true
+	}
+	// spec:
+	// "Two identifiers are different if they are spelled differently,
+	// or if they appear in different packages and are not exported.
+	// Otherwise, they are the same."
+	if obj.Name() != name {
+		return false
+	}
+	// obj.Name == name
+	if obj.Exported() {
+		return true
+	}
+	// not exported, so packages must be the same
+	if obj.Pkg() != nil && pkg != nil {
+		return obj.Pkg() == pkg
+	}
+	return obj.Pkg().Path() == pkg.Path()
+}
+
+// nify implements the core unification algorithm which is an
+// adapted version of Checker.identical. For changes to that
+// code the corresponding changes should be made here.
+// Must not be called directly from outside the unifier.
+func (u *unifier) nify(x, y types.Type, mode unifyMode) (result bool) {
+	u.depth++
+	defer func() {
+		u.depth--
+	}()
+
+	// nothing to do if x == y
+	if x == y || types.Unalias(x) == types.Unalias(y) {
+		return true
+	}
+
+	// Stop gap for cases where unification fails.
+	if u.depth > unificationDepthLimit {
+		if panicAtUnificationDepthLimit {
+			panic("unification reached recursion depth limit")
+		}
+		return false
+	}
+
+	// Unification is symmetric, so we can swap the operands.
+	// Ensure that if we have at least one
+	// - defined type, make sure one is in y
+	// - type parameter recorded with u, make sure one is in x
+	if asNamed(x) != nil || u.asBoundTypeParam(y) != nil {
+		x, y = y, x
+	}
+
+	// Unification will fail if we match a defined type against a type literal.
+	// If we are matching types in an assignment, at the top-level, types with
+	// the same type structure are permitted as long as at least one of them
+	// is not a defined type. To accommodate for that possibility, we continue
+	// unification with the underlying type of a defined type if the other type
+	// is a type literal. This is controlled by the exact unification mode.
+	// We also continue if the other type is a basic type because basic types
+	// are valid underlying types and may appear as core types of type constraints.
+	// If we exclude them, inferred defined types for type parameters may not
+	// match against the core types of their constraints (even though they might
+	// correctly match against some of the types in the constraint's type set).
+	// Finally, if unification (incorrectly) succeeds by matching the underlying
+	// type of a defined type against a basic type (because we include basic types
+	// as type literals here), and if that leads to an incorrectly inferred type,
+	// we will fail at function instantiation or argument assignment time.
+	//
+	// If we have at least one defined type, there is one in y.
+	if ny := asNamed(y); mode&unifyModeExact == 0 && ny != nil && isTypeLit(x) {
+		y = ny.Underlying()
+		// Per the spec, a defined type cannot have an underlying type
+		// that is a type parameter.
+		// x and y may be identical now
+		if x == y || types.Unalias(x) == types.Unalias(y) {
+			return true
+		}
+	}
+
+	// Cases where at least one of x or y is a type parameter recorded with u.
+	// If we have at least one type parameter, there is one in x.
+	// If we have exactly one type parameter, because it is in x,
+	// isTypeLit(x) is false and y was not changed above. In other
+	// words, if y was a defined type, it is still a defined type
+	// (relevant for the logic below).
+	switch px, py := u.asBoundTypeParam(x), u.asBoundTypeParam(y); {
+	case px != nil && py != nil:
+		// both x and y are type parameters
+		if u.join(px, py) {
+			return true
+		}
+		// both x and y have an inferred type - they must match
+		return u.nify(u.at(px), u.at(py), mode)
+
+	case px != nil:
+		// x is a type parameter, y is not
+		if x := u.at(px); x != nil {
+			// x has an inferred type which must match y
+			if u.nify(x, y, mode) {
+				// We have a match, possibly through underlying types.
+				xi := asInterface(x)
+				yi := asInterface(y)
+				xn := asNamed(x) != nil
+				yn := asNamed(y) != nil
+				// If we have two interfaces, what to do depends on
+				// whether they are named and their method sets.
+				if xi != nil && yi != nil {
+					// Both types are interfaces.
+					// If both types are defined types, they must be identical
+					// because unification doesn't know which type has the "right" name.
+					if xn && yn {
+						return types.Identical(x, y)
+					}
+					return false
+					// Below is the original code for reference
+
+					// In all other cases, the method sets must match.
+					// The types unified so we know that corresponding methods
+					// match and we can simply compare the number of methods.
+					// TODO(gri) We may be able to relax this rule and select
+					// the more general interface. But if one of them is a defined
+					// type, it's not clear how to choose and whether we introduce
+					// an order dependency or not. Requiring the same method set
+					// is conservative.
+					// if len(xi.typeSet().methods) != len(yi.typeSet().methods) {
+					// 	return false
+					// }
+				} else if xi != nil || yi != nil {
+					// One but not both of them are interfaces.
+					// In this case, either x or y could be viable matches for the corresponding
+					// type parameter, which means choosing either introduces an order dependence.
+					// Therefore, we must fail unification (go.dev/issue/60933).
+					return false
+				}
+				// If we have inexact unification and one of x or y is a defined type, select the
+				// defined type. This ensures that in a series of types, all matching against the
+				// same type parameter, we infer a defined type if there is one, independent of
+				// order. Type inference or assignment may fail, which is ok.
+				// Selecting a defined type, if any, ensures that we don't lose the type name;
+				// and since we have inexact unification, a value of equally named or matching
+				// undefined type remains assignable (go.dev/issue/43056).
+				//
+				// Similarly, if we have inexact unification and there are no defined types but
+				// channel types, select a directed channel, if any. This ensures that in a series
+				// of unnamed types, all matching against the same type parameter, we infer the
+				// directed channel if there is one, independent of order.
+				// Selecting a directional channel, if any, ensures that a value of another
+				// inexactly unifying channel type remains assignable (go.dev/issue/62157).
+				//
+				// If we have multiple defined channel types, they are either identical or we
+				// have assignment conflicts, so we can ignore directionality in this case.
+				//
+				// If we have defined and literal channel types, a defined type wins to avoid
+				// order dependencies.
+				if mode&unifyModeExact == 0 {
+					switch {
+					case xn:
+						// x is a defined type: nothing to do.
+					case yn:
+						// x is not a defined type and y is a defined type: select y.
+						u.set(px, y)
+					default:
+						// Neither x nor y are defined types.
+						if yc, _ := y.Underlying().(*types.Chan); yc != nil && yc.Dir() != types.SendRecv {
+							// y is a directed channel type: select y.
+							u.set(px, y)
+						}
+					}
+				}
+				return true
+			}
+			return false
+		}
+		// otherwise, infer type from y
+		u.set(px, y)
+		return true
+	}
+
+	// If u.EnableInterfaceInference is set and we don't require exact unification,
+	// if both types are interfaces, one interface must have a subset of the
+	// methods of the other and corresponding method signatures must unify.
+	// If only one type is an interface, all its methods must be present in the
+	// other type and corresponding method signatures must unify.
+
+	// Unless we have exact unification, neither x nor y are interfaces now.
+	// Except for unbound type parameters (see below), x and y must be structurally
+	// equivalent to unify.
+
+	// If we get here and x or y is a type parameter, they are unbound
+	// (not recorded with the unifier).
+	// Ensure that if we have at least one type parameter, it is in x
+	// (the earlier swap checks for _recorded_ type parameters only).
+	// This ensures that the switch switches on the type parameter.
+	//
+	// TODO(gri) Factor out type parameter handling from the switch.
+	if isTypeParam(y) {
+		x, y = y, x
+	}
+
+	// Type elements (array, slice, etc. elements) use emode for unification.
+	// Element types must match exactly if the types are used in an assignment.
+	emode := mode
+	if mode&unifyModeAssign != 0 {
+		emode |= unifyModeExact
+	}
+
+	// Continue with unaliased types but don't lose original alias names, if any (go.dev/issue/67628).
+	xorig, x := x, types.Unalias(x)
+	yorig, y := y, types.Unalias(y)
+
+	switch x := x.(type) {
+	case *types.Basic:
+		// Basic types are singletons except for the rune and byte
+		// aliases, thus we cannot solely rely on the x == y check
+		// above. See also comment in TypeName.IsAlias.
+		if y, ok := y.(*types.Basic); ok {
+			return x.Kind() == y.Kind()
+		}
+
+	case *types.Array:
+		// Two array types unify if they have the same array length
+		// and their element types unify.
+		if y, ok := y.(*types.Array); ok {
+			// If one or both array lengths are unknown (< 0) due to some error,
+			// assume they are the same to avoid spurious follow-on errors.
+			return (x.Len() < 0 || y.Len() < 0 || x.Len() == y.Len()) && u.nify(x.Elem(), y.Elem(), emode)
+		}
+
+	case *types.Slice:
+		// Two slice types unify if their element types unify.
+		if y, ok := y.(*types.Slice); ok {
+			return u.nify(x.Elem(), y.Elem(), emode)
+		}
+
+	case *types.Struct:
+		// Two struct types unify if they have the same sequence of fields,
+		// and if corresponding fields have the same names, their (field) types unify,
+		// and they have identical tags. Two embedded fields are considered to have the same
+		// name. Lower-case field names from different packages are always different.
+		if y, ok := y.(*types.Struct); ok {
+			if x.NumFields() == y.NumFields() {
+				for i := range x.NumFields() {
+					f := x.Field(i)
+					g := y.Field(i)
+					if f.Embedded() != g.Embedded() ||
+						x.Tag(i) != y.Tag(i) ||
+						!sameId(f, g.Pkg(), g.Name(), false) ||
+						!u.nify(f.Type(), g.Type(), emode) {
+						return false
+					}
+				}
+				return true
+			}
+		}
+
+	case *types.Pointer:
+		// Two pointer types unify if their base types unify.
+		if y, ok := y.(*types.Pointer); ok {
+			return u.nify(x.Elem(), y.Elem(), emode)
+		}
+
+	case *types.Tuple:
+		// Two tuples types unify if they have the same number of elements
+		// and the types of corresponding elements unify.
+		if y, ok := y.(*types.Tuple); ok {
+			if x.Len() == y.Len() {
+				if x != nil {
+					for i := range x.Len() {
+						v := x.At(i)
+						w := y.At(i)
+						if !u.nify(v.Type(), w.Type(), mode) {
+							return false
+						}
+					}
+				}
+				return true
+			}
+		}
+
+	case *types.Signature:
+		// Two function types unify if they have the same number of parameters
+		// and result values, corresponding parameter and result types unify,
+		// and either both functions are variadic or neither is.
+		// Parameter and result names are not required to match.
+		// TODO(gri) handle type parameters or document why we can ignore them.
+		if y, ok := y.(*types.Signature); ok {
+			return x.Variadic() == y.Variadic() &&
+				u.nify(x.Params(), y.Params(), emode) &&
+				u.nify(x.Results(), y.Results(), emode)
+		}
+
+	case *types.Interface:
+		return false
+		// Below is the original code
+
+		// Two interface types unify if they have the same set of methods with
+		// the same names, and corresponding function types unify.
+		// Lower-case method names from different packages are always different.
+		// The order of the methods is irrelevant.
+		// xset := x.typeSet()
+		// yset := y.typeSet()
+		// if xset.comparable != yset.comparable {
+		// 	return false
+		// }
+		// if !xset.terms.equal(yset.terms) {
+		// 	return false
+		// }
+		// a := xset.methods
+		// b := yset.methods
+		// if len(a) == len(b) {
+		// 	// Interface types are the only types where cycles can occur
+		// 	// that are not "terminated" via named types; and such cycles
+		// 	// can only be created via method parameter types that are
+		// 	// anonymous interfaces (directly or indirectly) embedding
+		// 	// the current interface. Example:
+		// 	//
+		// 	//    type T interface {
+		// 	//        m() interface{T}
+		// 	//    }
+		// 	//
+		// 	// If two such (differently named) interfaces are compared,
+		// 	// endless recursion occurs if the cycle is not detected.
+		// 	//
+		// 	// If x and y were compared before, they must be equal
+		// 	// (if they were not, the recursion would have stopped);
+		// 	// search the ifacePair stack for the same pair.
+		// 	//
+		// 	// This is a quadratic algorithm, but in practice these stacks
+		// 	// are extremely short (bounded by the nesting depth of interface
+		// 	// type declarations that recur via parameter types, an extremely
+		// 	// rare occurrence). An alternative implementation might use a
+		// 	// "visited" map, but that is probably less efficient overall.
+		// 	q := &ifacePair{x, y, p}
+		// 	for p != nil {
+		// 		if p.identical(q) {
+		// 			return true // same pair was compared before
+		// 		}
+		// 		p = p.prev
+		// 	}
+		// 	if debug {
+		// 		assertSortedMethods(a)
+		// 		assertSortedMethods(b)
+		// 	}
+		// 	for i, f := range a {
+		// 		g := b[i]
+		// 		if f.Id() != g.Id() || !u.nify(f.typ, g.typ, exact, q) {
+		// 			return false
+		// 		}
+		// 	}
+		// 	return true
+		// }
+
+	case *types.Map:
+		// Two map types unify if their key and value types unify.
+		if y, ok := y.(*types.Map); ok {
+			return u.nify(x.Key(), y.Key(), emode) && u.nify(x.Elem(), y.Elem(), emode)
+		}
+
+	case *types.Chan:
+		// Two channel types unify if their value types unify
+		// and if they have the same direction.
+		// The channel direction is ignored for inexact unification.
+		if y, ok := y.(*types.Chan); ok {
+			return (mode&unifyModeExact == 0 || x.Dir() == y.Dir()) && u.nify(x.Elem(), y.Elem(), emode)
+		}
+
+	case *types.Named:
+		// Two named types unify if their type names originate in the same type declaration.
+		// If they are instantiated, their type argument lists must unify.
+		if y := asNamed(y); y != nil {
+			// Check type arguments before origins so they unify
+			// even if the origins don't match; for better error
+			// messages (see go.dev/issue/53692).
+			xargs := x.TypeArgs()
+			yargs := y.TypeArgs()
+			if xargs.Len() != yargs.Len() {
+				return false
+			}
+			for i := range xargs.Len() {
+				xarg := xargs.At(i)
+				yarg := yargs.At(i)
+				if !u.nify(xarg, yarg, mode) {
+					return false
+				}
+			}
+			return identicalOrigin(x, y)
+		}
+
+	case *types.TypeParam:
+		// By definition, a valid type argument must be in the type set of
+		// the respective type constraint. Therefore, the type argument's
+		// underlying type must be in the set of underlying types of that
+		// constraint. If there is a single such underlying type, it's the
+		// constraint's core type. It must match the type argument's under-
+		// lying type, irrespective of whether the actual type argument,
+		// which may be a defined type, is actually in the type set (that
+		// will be determined at instantiation time).
+		// Thus, if we have the core type of an unbound type parameter,
+		// we know the structure of the possible types satisfying such
+		// parameters. Use that core type for further unification
+		// (see go.dev/issue/50755 for a test case).
+		if enableCoreTypeUnification {
+			// Because the core type is always an underlying type,
+			// unification will take care of matching against a
+			// defined or literal type automatically.
+			// If y is also an unbound type parameter, we will end
+			// up here again with x and y swapped, so we don't
+			// need to take care of that case separately.
+			if cx := coreType(x); cx != nil {
+				// If y is a defined type, it may not match against cx which
+				// is an underlying type (incl. int, string, etc.). Use assign
+				// mode here so that the unifier automatically takes under(y)
+				// if necessary.
+				return u.nify(cx, yorig, unifyModeAssign)
+			}
+		}
+		// x != y and there's nothing to do
+
+	case nil:
+		// avoid a crash in case of nil type
+
+	default:
+		panic(fmt.Sprintf("u.nify(%s, %s, %d)", xorig, yorig, mode))
+	}
+
+	return false
+}
diff --git a/gopls/internal/test/integration/completion/completion_test.go b/gopls/internal/test/integration/completion/completion_test.go
index c96e569f1ad..1f6eb2fe0fb 100644
--- a/gopls/internal/test/integration/completion/completion_test.go
+++ b/gopls/internal/test/integration/completion/completion_test.go
@@ -970,6 +970,275 @@ use ./missing/
 	})
 }
 
+const reverseInferenceSrcPrelude = `
+-- go.mod --
+module mod.com
+
+go 1.18
+-- a.go --
+package a
+
+type InterfaceA interface {
+	implA()
+}
+
+type InterfaceB interface {
+	implB()
+}
+
+
+type TypeA struct{}
+
+func (TypeA) implA() {}
+
+type TypeX string
+
+func (TypeX) implB() {}
+
+type TypeB struct{}
+
+func (TypeB) implB() {}
+
+type TypeC struct{} // should have no impact
+
+type Wrap[T any] struct {
+	inner *T
+}
+
+func NewWrap[T any](x T) Wrap[T] {
+	return Wrap[T]{inner: &x}
+}
+
+func DoubleWrap[T any, U any](t T, u U) (Wrap[T], Wrap[U]) {
+	return Wrap[T]{inner: &t}, Wrap[U]{inner: &u}
+}
+
+func IntWrap[T int32 | int64](x T) Wrap[T] {
+	return Wrap[T]{inner: &x}
+}
+
+var ia InterfaceA
+var ib InterfaceB
+
+var avar TypeA
+var bvar TypeB
+
+var i int
+var i32 int32
+var i64 int64
+`
+
+func TestReverseInferCompletion(t *testing.T) {
+	src := reverseInferenceSrcPrelude + `
+	func main() {
+		var _ Wrap[int64] = IntWrap()
+	}
+	`
+	Run(t, src, func(t *testing.T, env *Env) {
+		compl := env.RegexpSearch("a.go", `IntWrap\(()\)`)
+
+		env.OpenFile("a.go")
+		result := env.Completion(compl)
+
+		wantLabel := []string{"i64", "i", "i32", "int64()"}
+
+		// only check the prefix due to formatting differences with escaped characters
+		wantText := []string{"i64", "int64(i", "int64(i32", "int64("}
+
+		for i, item := range result.Items[:len(wantLabel)] {
+			if diff := cmp.Diff(wantLabel[i], item.Label); diff != "" {
+				t.Errorf("Completion: unexpected label mismatch (-want +got):\n%s", diff)
+			}
+
+			if insertText, ok := item.TextEdit.Value.(protocol.InsertReplaceEdit); ok {
+				if diff := cmp.Diff(wantText[i], insertText.NewText[:len(wantText[i])]); diff != "" {
+					t.Errorf("Completion: unexpected insertText mismatch (checks prefix only) (-want +got):\n%s", diff)
+				}
+			}
+		}
+	})
+}
+
+func TestInterfaceReverseInferCompletion(t *testing.T) {
+	src := reverseInferenceSrcPrelude + `
+	func main() {
+		var wa Wrap[InterfaceA]
+		var wb Wrap[InterfaceB]
+		wb = NewWrap() // wb is of type Wrap[InterfaceB]
+	}
+	`
+
+	Run(t, src, func(t *testing.T, env *Env) {
+		compl := env.RegexpSearch("a.go", `NewWrap\(()\)`)
+
+		env.OpenFile("a.go")
+		result := env.Completion(compl)
+
+		wantLabel := []string{"ib", "bvar", "wb.inner", "TypeB{}", "TypeX()", "nil"}
+
+		// only check the prefix due to formatting differences with escaped characters
+		wantText := []string{"ib", "InterfaceB(", "*wb.inner", "InterfaceB(", "InterfaceB(", "nil"}
+
+		for i, item := range result.Items[:len(wantLabel)] {
+			if diff := cmp.Diff(wantLabel[i], item.Label); diff != "" {
+				t.Errorf("Completion: unexpected label mismatch (-want +got):\n%s", diff)
+			}
+
+			if insertText, ok := item.TextEdit.Value.(protocol.InsertReplaceEdit); ok {
+				if diff := cmp.Diff(wantText[i], insertText.NewText[:len(wantText[i])]); diff != "" {
+					t.Errorf("Completion: unexpected insertText mismatch (checks prefix only) (-want +got):\n%s", diff)
+				}
+			}
+		}
+	})
+}
+
+func TestInvalidReverseInferenceDefaultsToConstraintCompletion(t *testing.T) {
+	src := reverseInferenceSrcPrelude + `
+	func main() {
+		var wa Wrap[InterfaceA]
+		// This is ambiguous, so default to the constraint rather the inference.
+		wa = IntWrap()
+	}
+	`
+	Run(t, src, func(t *testing.T, env *Env) {
+		compl := env.RegexpSearch("a.go", `IntWrap\(()\)`)
+
+		env.OpenFile("a.go")
+		result := env.Completion(compl)
+
+		wantLabel := []string{"i32", "i64", "nil"}
+
+		for i, item := range result.Items[:len(wantLabel)] {
+			if diff := cmp.Diff(wantLabel[i], item.Label); diff != "" {
+				t.Errorf("Completion: unexpected label mismatch (-want +got):\n%s", diff)
+			}
+		}
+	})
+}
+
+func TestInterfaceReverseInferTypeParamCompletion(t *testing.T) {
+	src := reverseInferenceSrcPrelude + `
+	func main() {
+		var wa Wrap[InterfaceA]
+		var wb Wrap[InterfaceB]
+		wb = NewWrap[]()
+	}
+	`
+
+	Run(t, src, func(t *testing.T, env *Env) {
+		compl := env.RegexpSearch("a.go", `NewWrap\[()\]\(\)`)
+
+		env.OpenFile("a.go")
+		result := env.Completion(compl)
+		want := []string{"InterfaceB", "TypeB", "TypeX", "InterfaceA", "TypeA"}
+		for i, item := range result.Items[:len(want)] {
+			if diff := cmp.Diff(want[i], item.Label); diff != "" {
+				t.Errorf("Completion: unexpected mismatch (-want +got):\n%s", diff)
+			}
+		}
+	})
+}
+
+func TestInvalidReverseInferenceTypeParamDefaultsToConstraintCompletion(t *testing.T) {
+	src := reverseInferenceSrcPrelude + `
+	func main() {
+		var wa Wrap[InterfaceA]
+		// This is ambiguous, so default to the constraint rather the inference.
+		wb = IntWrap[]()
+	}
+	`
+
+	Run(t, src, func(t *testing.T, env *Env) {
+		compl := env.RegexpSearch("a.go", `IntWrap\[()\]\(\)`)
+
+		env.OpenFile("a.go")
+		result := env.Completion(compl)
+		want := []string{"int32", "int64"}
+		for i, item := range result.Items[:len(want)] {
+			if diff := cmp.Diff(want[i], item.Label); diff != "" {
+				t.Errorf("Completion: unexpected mismatch (-want +got):\n%s", diff)
+			}
+		}
+	})
+}
+
+func TestReverseInferDoubleTypeParamCompletion(t *testing.T) {
+	src := reverseInferenceSrcPrelude + `
+	func main() {
+		var wa Wrap[InterfaceA]
+		var wb Wrap[InterfaceB]
+
+		wa, wb = DoubleWrap[]()
+		// _ is necessary to trick the parser into an index list expression
+		wa, wb = DoubleWrap[InterfaceA, _]()
+	}
+	`
+	Run(t, src, func(t *testing.T, env *Env) {
+		env.OpenFile("a.go")
+
+		compl := env.RegexpSearch("a.go", `DoubleWrap\[()\]\(\)`)
+		result := env.Completion(compl)
+
+		wantLabel := []string{"InterfaceA", "TypeA", "InterfaceB", "TypeB", "TypeC"}
+
+		for i, item := range result.Items[:len(wantLabel)] {
+			if diff := cmp.Diff(wantLabel[i], item.Label); diff != "" {
+				t.Errorf("Completion: unexpected label mismatch (-want +got):\n%s", diff)
+			}
+		}
+
+		compl = env.RegexpSearch("a.go", `DoubleWrap\[InterfaceA, (_)\]\(\)`)
+		result = env.Completion(compl)
+
+		wantLabel = []string{"InterfaceB", "TypeB", "TypeX", "InterfaceA", "TypeA"}
+
+		for i, item := range result.Items[:len(wantLabel)] {
+			if diff := cmp.Diff(wantLabel[i], item.Label); diff != "" {
+				t.Errorf("Completion: unexpected label mismatch (-want +got):\n%s", diff)
+			}
+		}
+	})
+}
+
+func TestDoubleParamReturnCompletion(t *testing.T) {
+	src := reverseInferenceSrcPrelude + `
+	func concrete() (Wrap[InterfaceA], Wrap[InterfaceB]) {
+		return DoubleWrap[]()
+	}
+
+	func concrete2() (Wrap[InterfaceA], Wrap[InterfaceB]) {
+		return DoubleWrap[InterfaceA, _]()
+	}
+	`
+
+	Run(t, src, func(t *testing.T, env *Env) {
+		env.OpenFile("a.go")
+
+		compl := env.RegexpSearch("a.go", `DoubleWrap\[()\]\(\)`)
+		result := env.Completion(compl)
+
+		wantLabel := []string{"InterfaceA", "TypeA", "InterfaceB", "TypeB", "TypeC"}
+
+		for i, item := range result.Items[:len(wantLabel)] {
+			if diff := cmp.Diff(wantLabel[i], item.Label); diff != "" {
+				t.Errorf("Completion: unexpected label mismatch (-want +got):\n%s", diff)
+			}
+		}
+
+		compl = env.RegexpSearch("a.go", `DoubleWrap\[InterfaceA, (_)\]\(\)`)
+		result = env.Completion(compl)
+
+		wantLabel = []string{"InterfaceB", "TypeB", "TypeX", "InterfaceA", "TypeA"}
+
+		for i, item := range result.Items[:len(wantLabel)] {
+			if diff := cmp.Diff(wantLabel[i], item.Label); diff != "" {
+				t.Errorf("Completion: unexpected label mismatch (-want +got):\n%s", diff)
+			}
+		}
+	})
+}
+
 func TestBuiltinCompletion(t *testing.T) {
 	const files = `
 -- go.mod --