highperfocused/ollama
1
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2025-11-12 11:47:19 +01:00
Code Issues Packages Projects Releases Wiki Activity

thinking: allow "think": false for non-thinking models (#12555)

Browse Source
This commit is contained in:
Patrick Devine
2025-10-09 18:46:00 -07:00
committed by GitHub
parent 47298fce39
commit d681cd7c29
2 changed files with 30 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
server/routes.go
View File

@@ -332,14 +332,16 @@ func (s *Server) GenerateHandler(c *gin.Context) {
} }
modelCaps := m.Capabilities() modelCaps := m.Capabilities()
if req.Think != nil {
caps = append(caps, model.CapabilityThinking)
} else {
// add thinking if the model supports it
if slices.Contains(modelCaps, model.CapabilityThinking) { if slices.Contains(modelCaps, model.CapabilityThinking) {
caps = append(caps, model.CapabilityThinking) caps = append(caps, model.CapabilityThinking)
if req.Think == nil {
req.Think = &api.ThinkValue{Value: true} req.Think = &api.ThinkValue{Value: true}
} }
} else {
if req.Think != nil && req.Think.Bool() {
c.JSON(http.StatusBadRequest, gin.H{"error": fmt.Sprintf("%q does not support thinking", req.Model)})
return
}
} }
r, m, opts, err := s.scheduleRunner(c.Request.Context(), name.String(), caps, req.Options, req.KeepAlive) r, m, opts, err := s.scheduleRunner(c.Request.Context(), name.String(), caps, req.Options, req.KeepAlive)
@@ -1877,14 +1879,16 @@ func (s *Server) ChatHandler(c *gin.Context) {
} }
modelCaps := m.Capabilities() modelCaps := m.Capabilities()
if req.Think != nil {
caps = append(caps, model.CapabilityThinking)
} else {
// add thinking if the model supports it
if slices.Contains(modelCaps, model.CapabilityThinking) { if slices.Contains(modelCaps, model.CapabilityThinking) {
caps = append(caps, model.CapabilityThinking) caps = append(caps, model.CapabilityThinking)
if req.Think == nil {
req.Think = &api.ThinkValue{Value: true} req.Think = &api.ThinkValue{Value: true}
} }
} else {
if req.Think != nil && req.Think.Bool() {
c.JSON(http.StatusBadRequest, gin.H{"error": fmt.Sprintf("%q does not support thinking", req.Model)})
return
}
} }
r, m, opts, err := s.scheduleRunner(c.Request.Context(), name.String(), caps, req.Options, req.KeepAlive) r, m, opts, err := s.scheduleRunner(c.Request.Context(), name.String(), caps, req.Options, req.KeepAlive)

17
server/routes_generate_test.go
View File

@@ -158,11 +158,26 @@ func TestGenerateChat(t *testing.T) {
t.Errorf("expected status 400, got %d", w.Code) t.Errorf("expected status 400, got %d", w.Code)
} }
if diff := cmp.Diff(w.Body.String(), `{"error":"registry.ollama.ai/library/test:latest does not support thinking"}`); diff != "" { if diff := cmp.Diff(w.Body.String(), `{"error":"\"test\" does not support thinking"}`); diff != "" {
t.Errorf("mismatch (-got +want):\n%s", diff) t.Errorf("mismatch (-got +want):\n%s", diff)
} }
}) })
t.Run("model can't think but think set false", func(t *testing.T) {
think := false
w := createRequest(t, s.ChatHandler, api.ChatRequest{
Model: "test",
Messages: []api.Message{
{Role: "user", Content: "Hello!"},
},
Think: &api.ThinkValue{Value: think},
})
if w.Code != http.StatusOK {
t.Errorf("expected status 200, got %d", w.Code)
}
})
t.Run("missing model", func(t *testing.T) { t.Run("missing model", func(t *testing.T) {
w := createRequest(t, s.ChatHandler, api.ChatRequest{}) w := createRequest(t, s.ChatHandler, api.ChatRequest{})
if w.Code != http.StatusBadRequest { if w.Code != http.StatusBadRequest {