Skip to content

Open source AI coding agent. Designed for large projects and real world tasks.

updated model names for gemini and others, added assortment of built-in models for ollama (some with openrouter options as well), ollama experimental and adaptive built-in model packs, some fixes for ollama, and some docs updates

Dane Schneidere2ab45514d1a
diff --git a/app/cli/cmd/models.go b/app/cli/cmd/models.go
index 0813e590..9f5b52ff 100644
--- a/app/cli/cmd/models.go
+++ b/app/cli/cmd/models.go
@@ -131,7 +131,7 @@ func manageCustomModels(cmd *cobra.Command, args []string) {
usingDefaultPath := false
if customModelsPath == "" {
usingDefaultPath = true
- customModelsPath = lib.CustomModelsDefaultPath
+ customModelsPath = lib.GetCustomModelsPath(auth.Current.UserId)
}
exists, err := fs.FileExists(customModelsPath)
@@ -568,15 +568,39 @@ func getExampleTemplate(isCloud, isCloudIntegratedModels bool) shared.ClientMode
Name: "example-model-pack",
Description: "Example model pack",
ClientModelPackSchemaRoles: shared.ClientModelPackSchemaRoles{
- Planner: "deepseek/r1",
- Architect: "deepseek/r1",
- Coder: "deepseek/v3-0324",
- PlanSummary: lightModelId,
- Builder: "deepseek/r1-hidden",
- WholeFileBuilder: "deepseek/r1-hidden",
- ExecStatus: "deepseek/r1-hidden",
- Namer: lightModelId,
- CommitMsg: lightModelId,
+ Planner: "deepseek/r1",
+ Architect: "deepseek/r1",
+ Coder: &shared.ModelRoleConfigSchema{
+ ModelId: "deepseek/v3",
+ LargeContextFallback: &shared.ModelRoleConfigSchema{
+ ModelId: "google/gemini-2.5-pro",
+ },
+ ErrorFallback: &shared.ModelRoleConfigSchema{
+ ModelId: "deepseek/r1-hidden",
+ },
+ },
+ PlanSummary: lightModelId,
+ Builder: shared.ModelRoleConfigSchema{
+ ModelId: "deepseek/r1-hidden",
+ StrongModel: &shared.ModelRoleConfigSchema{
+ ModelId: "openai/o3-medium",
+ },
+ },
+ WholeFileBuilder: shared.ModelRoleConfigSchema{
+ ModelId: "deepseek/r1-hidden",
+ LargeContextFallback: &shared.ModelRoleConfigSchema{
+ ModelId: "google/gemini-2.5-pro",
+ LargeOutputFallback: &shared.ModelRoleConfigSchema{
+ ModelId: "openai/o3-low",
+ },
+ },
+ LargeOutputFallback: &shared.ModelRoleConfigSchema{
+ ModelId: "openai/o3-low",
+ },
+ },
+ ExecStatus: "deepseek/r1-hidden",
+ Namer: lightModelId,
+ CommitMsg: lightModelId,
},
},
},
diff --git a/app/cli/cmd/plan_start_helpers.go b/app/cli/cmd/plan_start_helpers.go
index 92a5b024..264de1fd 100644
--- a/app/cli/cmd/plan_start_helpers.go
+++ b/app/cli/cmd/plan_start_helpers.go
@@ -21,12 +21,12 @@ var (
fullAuto bool
// Type flags
- dailyModels bool
- reasoningModels bool
- strongModels bool
- ossModels bool
- cheapModels bool
- geminiPreviewModels bool
+ dailyModels bool
+ reasoningModels bool
+ strongModels bool
+ ossModels bool
+ cheapModels bool
+ geminiModels bool
)
func AddNewPlanFlags(cmd *cobra.Command) {
@@ -43,7 +43,7 @@ func AddNewPlanFlags(cmd *cobra.Command) {
cmd.Flags().BoolVar(&strongModels, "strong", false, shared.StrongModelPack.Description)
cmd.Flags().BoolVar(&cheapModels, "cheap", false, shared.CheapModelPack.Description)
cmd.Flags().BoolVar(&ossModels, "oss", false, shared.OSSModelPack.Description)
- cmd.Flags().BoolVar(&geminiPreviewModels, "gemini-preview", false, shared.GeminiPreviewModelPack.Description)
+ cmd.Flags().BoolVar(&geminiModels, "gemini", false, shared.GeminiModelPack.Description)
}
func resolveAutoMode(config *shared.PlanConfig) (bool, *shared.PlanConfig) {
@@ -139,8 +139,8 @@ func resolveModelPackWithArgs(settings *shared.PlanSettings, silent bool) (*shar
packName = shared.ReasoningModelPack.Name
} else if dailyModels {
packName = shared.DailyDriverModelPack.Name
- } else if geminiPreviewModels {
- packName = shared.GeminiPreviewModelPack.Name
+ } else if geminiModels {
+ packName = shared.GeminiModelPack.Name
}
if packName != "" && packName != originalSettings.GetModelPack().Name {
diff --git a/app/cli/cmd/repl.go b/app/cli/cmd/repl.go
index d1157dec..e41c8bca 100644
--- a/app/cli/cmd/repl.go
+++ b/app/cli/cmd/repl.go
@@ -130,7 +130,7 @@ func runRepl(cmd *cobra.Command, args []string) {
args = append(args, "--daily")
} else if reasoningModels {
args = append(args, "--reasoning")
- } else if geminiPreviewModels {
+ } else if geminiModels {
args = append(args, "--gemini-preview")
}
diff --git a/app/cli/lib/custom_models.go b/app/cli/lib/custom_models.go
index d071bef9..a13f2759 100644
--- a/app/cli/lib/custom_models.go
+++ b/app/cli/lib/custom_models.go
@@ -16,15 +16,13 @@ import (
"github.com/fatih/color"
)
-var CustomModelsDefaultPath string
-
type CustomModelsCheckLocalChangesResult struct {
HasLocalChanges bool
LocalModelsInput shared.ModelsInput
}
-func init() {
- CustomModelsDefaultPath = filepath.Join(fs.HomePlandexDir, "custom-models.json")
+func GetCustomModelsPath(userId string) string {
+ return filepath.Join(fs.HomePlandexDir, "accounts", userId, "custom-models.json")
}
func GetServerModelsInput() (*shared.ModelsInput, error) {
@@ -330,12 +328,17 @@ func MustSyncCustomModels(path string, serverModelsInput *shared.ModelsInput) bo
}
func SyncCustomModels() error {
+ userId := auth.Current.UserId
+ if userId == "" {
+ return fmt.Errorf("auth.Current.UserId is empty")
+ }
+
serverModelsInput, err := GetServerModelsInput()
if err != nil {
return fmt.Errorf("error getting server models input: %v", err)
}
- MustSyncCustomModels(CustomModelsDefaultPath, serverModelsInput)
+ MustSyncCustomModels(GetCustomModelsPath(userId), serverModelsInput)
return nil
}
diff --git a/app/cli/lib/models_sync.go b/app/cli/lib/models_sync.go
index 491a099e..a781a96f 100644
--- a/app/cli/lib/models_sync.go
+++ b/app/cli/lib/models_sync.go
@@ -2,6 +2,7 @@ package lib
import (
"fmt"
+ "plandex-cli/auth"
"plandex-cli/term"
"github.com/fatih/color"
@@ -11,7 +12,14 @@ func PromptSyncModelsIfNeeded() error {
var changes []string
var onApprove []func() error
- customModelsRes, err := CustomModelsCheckLocalChanges(CustomModelsDefaultPath)
+ userId := auth.Current.UserId
+ if userId == "" {
+ return fmt.Errorf("auth.Current.UserId is empty")
+ }
+
+ customModelsPath := GetCustomModelsPath(userId)
+
+ customModelsRes, err := CustomModelsCheckLocalChanges(customModelsPath)
if err != nil {
return fmt.Errorf("error checking custom models: %v", err)
}
@@ -19,7 +27,7 @@ func PromptSyncModelsIfNeeded() error {
if customModelsRes.HasLocalChanges {
changes = append(
changes,
- fmt.Sprintf("%s → %s", color.New(term.ColorHiCyan, color.Bold).Sprint("Custom models"), CustomModelsDefaultPath))
+ fmt.Sprintf("%s → %s", color.New(term.ColorHiCyan, color.Bold).Sprint("Custom models"), customModelsPath))
onApprove = append(onApprove, SyncCustomModels)
}
diff --git a/app/cli/term/help.go b/app/cli/term/help.go
index 1a49fb7b..1d74f768 100644
--- a/app/cli/term/help.go
+++ b/app/cli/term/help.go
@@ -43,8 +43,7 @@ var CliCommands = []CmdConfig{
{"new --strong", "", fmt.Sprintf("start a new plan with %s model pack", "'strong'"), true},
{"new --cheap", "", fmt.Sprintf("start a new plan with %s model pack", "'cheap'"), true},
{"new --oss", "", fmt.Sprintf("start a new plan with %s model pack", "'oss'"), true},
- {"new --gemini-preview", "", fmt.Sprintf("start a new plan with %s model pack", "'gemini-preview'"), true},
- // {"new --crazy", "", fmt.Sprintf("start a new plan with %s model pack", "'crazy'"), true},
+ {"new --gemini", "", fmt.Sprintf("start a new plan with %s model pack", "'gemini'"), true},
{"plans", "pl", "list plans", true},
{"cd", "", "set current plan by name or index", true},
{"current", "cu", "show current plan", true},
@@ -127,7 +126,7 @@ var CliCommands = []CmdConfig{
{"set-model strong", "", fmt.Sprintf("Use %s model pack", "'strong'"), true},
{"set-model cheap", "", fmt.Sprintf("Use %s model pack", "'cheap'"), true},
{"set-model oss", "", fmt.Sprintf("Use %s model pack", "'oss'"), true},
- {"set-model gemini-preview", "", fmt.Sprintf("Use %s model pack", "'gemini-preview'"), true},
+ {"set-model gemini", "", fmt.Sprintf("Use %s model pack", "'gemini'"), true},
{"ps", "", "list active and recently finished plan streams", true},
{"stop", "", "stop an active plan stream", true},
@@ -342,7 +341,7 @@ func PrintHelpAllCommands() {
fmt.Fprintln(builder)
color.New(color.Bold, color.BgCyan, color.FgHiWhite).Fprintln(builder, " New Plan Shortcuts ")
- printCmds(builder, " ", []color.Attribute{color.Bold, ColorHiCyan}, "new --full", "new --semi", "new --plus", "new --basic", "new --none", "new --daily", "new --reasoning", "new --strong", "new --cheap", "new --oss", "new --gemini-preview" /*"new --crazy"*/)
+ printCmds(builder, " ", []color.Attribute{color.Bold, ColorHiCyan}, "new --full", "new --semi", "new --plus", "new --basic", "new --none", "new --daily", "new --reasoning", "new --strong", "new --cheap", "new --oss", "new --gemini" /*"new --crazy"*/)
fmt.Fprintln(builder)
fmt.Print(builder.String())
diff --git a/app/docker-compose.yml b/app/docker-compose.yml
index 554682b3..313fcfed 100644
--- a/app/docker-compose.yml
+++ b/app/docker-compose.yml
@@ -27,6 +27,10 @@ services:
GOENV: development
LOCAL_MODE: 1
PLANDEX_BASE_DIR: /plandex-server
+ OLLAMA_BASE_URL: http://host.docker.internal:11434
+
+ extra_hosts:
+ - "host.docker.internal:host-gateway"
networks:
- plandex-network
depends_on:
diff --git a/app/server/handlers/models.go b/app/server/handlers/models.go
index f309d705..476a071b 100644
--- a/app/server/handlers/models.go
+++ b/app/server/handlers/models.go
@@ -31,7 +31,7 @@ func UpsertCustomModelsHandler(w http.ResponseWriter, r *http.Request) {
var modelsInput shared.ModelsInput
if err := json.NewDecoder(r.Body).Decode(&modelsInput); err != nil {
log.Printf("Error decoding request body: %v\n", err)
- http.Error(w, "Invalid request body", http.StatusBadRequest)
+ http.Error(w, "Invalid request body: "+err.Error(), http.StatusBadRequest)
return
}
@@ -184,13 +184,13 @@ func UpsertCustomModelsHandler(w http.ResponseWriter, r *http.Request) {
}
for _, model := range updatedModelsInput.CustomModels {
- // ensure that providers to upsert are either built-in, being created, or already exist
+ // ensure that providers to upsert are either built-in, being imported, or already exist
for _, provider := range model.Providers {
if provider.Provider == shared.ModelProviderCustom {
_, exists := existingCustomProviderNames[*provider.CustomProvider]
_, creating := inputProviderNames[*provider.CustomProvider]
if !exists && !creating {
- msg := fmt.Sprintf("'%s' is not a custom model provider that exists or is being created", *provider.CustomProvider)
+ msg := fmt.Sprintf("'%s' is not a custom model provider that exists or is being imported", *provider.CustomProvider)
log.Println(msg)
http.Error(w, msg, http.StatusUnprocessableEntity)
return
@@ -214,7 +214,7 @@ func UpsertCustomModelsHandler(w http.ResponseWriter, r *http.Request) {
}
for _, modelPack := range updatedModelsInput.CustomModelPacks {
- // ensure that all models are either built-in, being created, or already exist
+ // ensure that all models are either built-in, being imported, or already exist
allModelIds := modelPack.AllModelIds()
for _, modelId := range allModelIds {
@@ -223,7 +223,7 @@ func UpsertCustomModelsHandler(w http.ResponseWriter, r *http.Request) {
_, builtIn := shared.BuiltInBaseModelsById[modelId]
if !exists && !creating && !builtIn {
- msg := fmt.Sprintf("'%s' is not built-in, not being created, and not an existing custom model", modelId)
+ msg := fmt.Sprintf("'%s' is not built-in, not being imported, and not an existing custom model", modelId)
log.Println(msg)
http.Error(w, msg, http.StatusUnprocessableEntity)
return
diff --git a/app/server/model/client.go b/app/server/model/client.go
index 1ba3d8ba..cf026841 100644
--- a/app/server/model/client.go
+++ b/app/server/model/client.go
@@ -222,7 +222,7 @@ func createChatCompletionStreamExtended(
}
case shared.ModelProviderAzureOpenAI:
if authVars["AZURE_API_BASE"] != "" {
- extendedReq.AzureApiBase = authVars["AZURE_API_BASE"]
+ extendedReq.LiteLLMApiBase = authVars["AZURE_API_BASE"]
}
if authVars["AZURE_API_VERSION"] != "" {
extendedReq.AzureApiVersion = authVars["AZURE_API_VERSION"]
@@ -266,6 +266,11 @@ func createChatCompletionStreamExtended(
if authVars["AWS_INFERENCE_PROFILE_ARN"] != "" {
extendedReq.BedrockInferenceProfileArn = authVars["AWS_INFERENCE_PROFILE_ARN"]
}
+
+ case shared.ModelProviderOllama:
+ if os.Getenv("OLLAMA_BASE_URL") != "" {
+ extendedReq.LiteLLMApiBase = os.Getenv("OLLAMA_BASE_URL")
+ }
}
// Marshal the request body to JSON
@@ -478,6 +483,7 @@ func resolveReq(req *types.ExtendedChatCompletionRequest, modelConfig *shared.Mo
log.Println("Ollama - clearing temperature and top p")
req.Temperature = 0
req.TopP = 0
+
}
}
diff --git a/app/server/model/litellm.go b/app/server/model/litellm.go
index bc26619c..93a9bd36 100644
--- a/app/server/model/litellm.go
+++ b/app/server/model/litellm.go
@@ -122,6 +122,12 @@ func startLiteLLMServer(numWorkers int) error {
"HOME=" + os.Getenv("HOME"),
}
+ if os.Getenv("OLLAMA_BASE_URL") != "" {
+ log.Println("OLLAMA_BASE_URL is set, so we can reach ollama from inside docker container in local mode")
+ // so we can reach ollama from inside docker container in local mode
+ liteLLMCmd.Env = append(liteLLMCmd.Env, "OLLAMA_BASE_URL="+os.Getenv("OLLAMA_BASE_URL"))
+ }
+
liteLLMCmd.Stdout = os.Stdout
liteLLMCmd.Stderr = os.Stderr
diff --git a/app/server/model/plan/tell_exec.go b/app/server/model/plan/tell_exec.go
index 10f3816c..89f5fa69 100644
--- a/app/server/model/plan/tell_exec.go
+++ b/app/server/model/plan/tell_exec.go
@@ -408,11 +408,23 @@ func execTellPlan(params execTellPlanParams) {
log.Println("Tell plan - got modelConfig for implementation stage")
}
- // log.Println("Tell plan - modelConfig:", spew.Sdump(modelConfig))
state.modelConfig = &modelConfig
baseModelConfig := modelConfig.GetBaseModelConfig(authVars, state.settings)
+ if baseModelConfig == nil {
+ log.Println("Tell plan - baseModelConfig is nil")
+ log.Println("Tell plan - modelConfig id:", modelConfig.ModelId)
+
+ go notify.NotifyErr(notify.SeverityError, fmt.Errorf("No model config found for: %s", state.modelConfig.ModelId))
+ active.StreamDoneCh <- &shared.ApiError{
+ Type: shared.ApiErrorTypeOther,
+ Status: http.StatusInternalServerError,
+ Msg: "No model config found for: " + string(state.modelConfig.ModelId),
+ }
+ return
+ }
+
state.baseModelConfig = baseModelConfig
// if the model doesn't support cache control, remove the cache control spec from the messages
diff --git a/app/server/types/message.go b/app/server/types/message.go
index fbb953ad..d2037075 100644
--- a/app/server/types/message.go
+++ b/app/server/types/message.go
@@ -130,6 +130,9 @@ type ExtendedChatCompletionRequest struct {
Prediction *OpenAIPrediction `json:"prediction,omitempty"`
Provider *OpenRouterProviderConfig `json:"provider,omitempty"`
+ // LiteLLM api base
+ LiteLLMApiBase string `json:"api_base,omitempty"`
+
// Openrouter/LiteLLM reasoning
ReasoningConfig *ReasoningConfig `json:"reasoning,omitempty"`
@@ -139,7 +142,6 @@ type ExtendedChatCompletionRequest struct {
VertexCredentials string `json:"vertex_credentials,omitempty"`
// Azure OpenAI request vars
- AzureApiBase string `json:"api_base,omitempty"`
AzureApiVersion string `json:"api_version,omitempty"`
AzureReasoningEffort shared.ReasoningEffort `json:"reasoning_effort,omitempty"`
diff --git a/app/shared/ai_models_available.go b/app/shared/ai_models_available.go
index 5b05dc60..8ad074a6 100644
--- a/app/shared/ai_models_available.go
+++ b/app/shared/ai_models_available.go
@@ -234,7 +234,7 @@ var BuiltInModels = []*BaseModelConfigSchema{
},
Providers: []BaseModelUsesProvider{
{Provider: ModelProviderAnthropic, ModelName: "anthropic/claude-3-5-sonnet-latest"},
- {Provider: ModelProviderGoogleVertex, ModelName: "vertex_ai/claude-3-5-sonnet@20240620"},
+ {Provider: ModelProviderGoogleVertex, ModelName: "vertex_ai/claude-3-5-sonnet-v2@20241022"},
{Provider: ModelProviderAmazonBedrock, ModelName: "bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0"},
{Provider: ModelProviderOpenRouter, ModelName: "anthropic/claude-3.5-sonnet"},
},
@@ -267,44 +267,29 @@ var BuiltInModels = []*BaseModelConfigSchema{
},
Providers: []BaseModelUsesProvider{
{Provider: ModelProviderGoogleAIStudio, ModelName: "gemini/gemini-1.5-pro"},
- {Provider: ModelProviderGoogleVertex, ModelName: "vertex_ai/gemini-1.5-pro-latest"},
+ {Provider: ModelProviderGoogleVertex, ModelName: "vertex_ai/gemini-1.5-pro"},
{Provider: ModelProviderOpenRouter, ModelName: "google/gemini-pro-1.5"},
},
},
{
- ModelTag: "google/gemini-2.5-pro-preview",
+ ModelTag: "google/gemini-2.5-pro",
Publisher: ModelPublisherGoogle,
- Description: "Google Gemini 2.5 Pro (Preview)",
+ Description: "Google Gemini 2.5 Pro",
BaseModelShared: BaseModelShared{
DefaultMaxConvoTokens: 75000, MaxTokens: 1048576,
MaxOutputTokens: 65535, ReservedOutputTokens: 65535,
PreferredOutputFormat: ModelOutputFormatXml,
},
Providers: []BaseModelUsesProvider{
- {Provider: ModelProviderGoogleAIStudio, ModelName: "gemini/gemini-2.5-pro-preview-05-06"},
- {Provider: ModelProviderGoogleVertex, ModelName: "vertex_ai/gemini-2.5-pro-preview-05-06"},
- {Provider: ModelProviderOpenRouter, ModelName: "google/gemini-2.5-pro-preview"},
+ {Provider: ModelProviderGoogleAIStudio, ModelName: "gemini/gemini-2.5-pro"},
+ {Provider: ModelProviderGoogleVertex, ModelName: "vertex_ai/gemini-2.5-pro"},
+ {Provider: ModelProviderOpenRouter, ModelName: "google/gemini-2.5-pro"},
},
},
{
- ModelTag: "google/gemini-2.5-pro-exp",
+ ModelTag: "google/gemini-2.5-flash",
Publisher: ModelPublisherGoogle,
- Description: "Google Gemini 2.5 Pro (Experimental)",
- BaseModelShared: BaseModelShared{
- DefaultMaxConvoTokens: 75000, MaxTokens: 1048576,
- MaxOutputTokens: 65535, ReservedOutputTokens: 65535,
- PreferredOutputFormat: ModelOutputFormatXml,
- },
- Providers: []BaseModelUsesProvider{
- {Provider: ModelProviderGoogleAIStudio, ModelName: "gemini/gemini-2.5-pro-exp-03-25"},
- {Provider: ModelProviderGoogleVertex, ModelName: "vertex_ai/gemini-2.5-pro-exp-03-25"},
- {Provider: ModelProviderOpenRouter, ModelName: "google/gemini-2.5-pro-exp-03-25"},
- },
- },
- {
- ModelTag: "google/gemini-2.5-flash-preview",
- Publisher: ModelPublisherGoogle,
- Description: "Google Gemini 2.5 Flash (Preview)",
+ Description: "Google Gemini 2.5 Flash",
BaseModelShared: BaseModelShared{
DefaultMaxConvoTokens: 75000, MaxTokens: 1048576,
MaxOutputTokens: 65535, ReservedOutputTokens: 65535,
@@ -315,15 +300,15 @@ var BuiltInModels = []*BaseModelConfigSchema{
{VariantTag: "thinking", Description: "thinking", Overrides: BaseModelShared{IncludeReasoning: true}},
},
Providers: []BaseModelUsesProvider{
- {Provider: ModelProviderGoogleAIStudio, ModelName: "gemini/gemini-2.5-flash-preview-05-20"},
- {Provider: ModelProviderGoogleVertex, ModelName: "vertex_ai/gemini-2.5-flash-preview-05-20"},
- {Provider: ModelProviderOpenRouter, ModelName: "google/gemini-2.5-flash-preview-05-20"},
+ {Provider: ModelProviderGoogleAIStudio, ModelName: "gemini/gemini-2.5-flash"},
+ {Provider: ModelProviderGoogleVertex, ModelName: "vertex_ai/gemini-2.5-flash"},
+ {Provider: ModelProviderOpenRouter, ModelName: "google/gemini-2.5-flash"},
},
},
{
- ModelTag: "deepseek/v3-0324",
+ ModelTag: "deepseek/v3",
Publisher: ModelPublisherDeepSeek,
- Description: "DeepSeek V3 (0324)",
+ Description: "DeepSeek V3",
BaseModelShared: BaseModelShared{
DefaultMaxConvoTokens: 7500, MaxTokens: 64000,
MaxOutputTokens: 8192, ReservedOutputTokens: 8192,
@@ -331,7 +316,7 @@ var BuiltInModels = []*BaseModelConfigSchema{
},
Providers: []BaseModelUsesProvider{
{Provider: ModelProviderDeepSeek, ModelName: "deepseek/deepseek-chat"},
- {Provider: ModelProviderOpenRouter, ModelName: "deepseek/deepseek-chat-v3-0324"},
+ {Provider: ModelProviderOpenRouter, ModelName: "deepseek/deepseek-chat-v3"},
},
},
{
@@ -339,8 +324,8 @@ var BuiltInModels = []*BaseModelConfigSchema{
Publisher: ModelPublisherDeepSeek,
Description: "DeepSeek R1",
BaseModelShared: BaseModelShared{
- DefaultMaxConvoTokens: 7500, MaxTokens: 64000,
- MaxOutputTokens: 8192, ReservedOutputTokens: 8192,
+ DefaultMaxConvoTokens: 7500, MaxTokens: 164000,
+ MaxOutputTokens: 33000, ReservedOutputTokens: 20000,
PreferredOutputFormat: ModelOutputFormatXml,
},
Variants: []BaseModelConfigVariant{
@@ -349,7 +334,59 @@ var BuiltInModels = []*BaseModelConfigSchema{
},
Providers: []BaseModelUsesProvider{
{Provider: ModelProviderDeepSeek, ModelName: "deepseek/deepseek-reasoner"},
- {Provider: ModelProviderOpenRouter, ModelName: "deepseek/deepseek-r1"},
+ {Provider: ModelProviderOpenRouter, ModelName: "deepseek/deepseek-r1-0528"},
+ },
+ },
+ {
+ ModelTag: "deepseek/r1-70b",
+ Publisher: ModelPublisherDeepSeek,
+ Description: "DeepSeek R1 70B",
+ BaseModelShared: BaseModelShared{
+ DefaultMaxConvoTokens: 10000, MaxTokens: 131072,
+ MaxOutputTokens: 131072, ReservedOutputTokens: 20000,
+ PreferredOutputFormat: ModelOutputFormatXml,
+ },
+ Providers: []BaseModelUsesProvider{
+ {Provider: ModelProviderOllama, ModelName: "ollama_chat/deepseek-r1:70b"},
+ },
+ },
+ {
+ ModelTag: "deepseek/r1-32b",
+ Publisher: ModelPublisherDeepSeek,
+ Description: "DeepSeek R1 32B",
+ BaseModelShared: BaseModelShared{
+ DefaultMaxConvoTokens: 10000, MaxTokens: 131072,
+ MaxOutputTokens: 131072, ReservedOutputTokens: 20000,
+ PreferredOutputFormat: ModelOutputFormatXml,
+ },
+ Providers: []BaseModelUsesProvider{
+ {Provider: ModelProviderOllama, ModelName: "ollama_chat/deepseek-r1:32b"},
+ },
+ },
+ {
+ ModelTag: "deepseek/r1-14b",
+ Publisher: ModelPublisherDeepSeek,
+ Description: "DeepSeek R1 14B",
+ BaseModelShared: BaseModelShared{
+ DefaultMaxConvoTokens: 10000, MaxTokens: 131072,
+ MaxOutputTokens: 131072, ReservedOutputTokens: 20000,
+ PreferredOutputFormat: ModelOutputFormatXml,
+ },
+ Providers: []BaseModelUsesProvider{
+ {Provider: ModelProviderOllama, ModelName: "ollama_chat/deepseek-r1:14b"},
+ },
+ },
+ {
+ ModelTag: "deepseek/r1-8b",
+ Publisher: ModelPublisherDeepSeek,
+ Description: "DeepSeek R1 8B",
+ BaseModelShared: BaseModelShared{
+ DefaultMaxConvoTokens: 10000, MaxTokens: 131072,
+ MaxOutputTokens: 131072, ReservedOutputTokens: 20000,
+ PreferredOutputFormat: ModelOutputFormatXml,
+ },
+ Providers: []BaseModelUsesProvider{
+ {Provider: ModelProviderOllama, ModelName: "ollama_chat/deepseek-r1:8b"},
},
},
{
@@ -370,6 +407,21 @@ var BuiltInModels = []*BaseModelConfigSchema{
{Provider: ModelProviderOpenRouter, ModelName: "perplexity/r1-1776"},
},
},
+
+ {
+ ModelTag: "perplexity/r1-1776-70b",
+ Publisher: ModelPublisherPerplexity,
+ Description: "Perplexity R1-1776 70B",
+ BaseModelShared: BaseModelShared{
+ DefaultMaxConvoTokens: 10000, MaxTokens: 131072,
+ MaxOutputTokens: 131072, ReservedOutputTokens: 20000,
+ PreferredOutputFormat: ModelOutputFormatXml,
+ },
+ Providers: []BaseModelUsesProvider{
+ {Provider: ModelProviderOllama, ModelName: "ollama_chat/r1-1776:70b"},
+ },
+ },
+
{
ModelTag: "perplexity/sonar-reasoning",
Publisher: ModelPublisherPerplexity,
diff --git a/app/shared/ai_models_data_models.go b/app/shared/ai_models_data_models.go
index b163053b..a5312d67 100644
--- a/app/shared/ai_models_data_models.go
+++ b/app/shared/ai_models_data_models.go
@@ -464,6 +464,33 @@ func (m *ModelRoleConfigSchema) toModelRoleConfig(role ModelRole) ModelRoleConfi
}
}
+func (m *ModelRoleConfigSchema) MarshalJSON() ([]byte, error) {
+ if m == nil {
+ return []byte("null"), nil
+ }
+ if m.bareRole() {
+ if m.ModelId == "" {
+ return []byte("null"), nil
+ }
+
+ return json.Marshal(string(m.ModelId)) // compact form
+ }
+ type alias ModelRoleConfigSchema
+ return json.Marshal((*alias)(m)) // full object
+}
+
+func (m *ModelRoleConfigSchema) UnmarshalJSON(data []byte) error {
+ // attempt the short string first
+ var s string
+ if err := json.Unmarshal(data, &s); err == nil {
+ *m = ModelRoleConfigSchema{ModelId: ModelId(s)}
+ return nil
+ }
+ // fallback to full object
+ type alias ModelRoleConfigSchema
+ return json.Unmarshal(data, (*alias)(m))
+}
+
func (m *ModelRoleConfig) ToModelRoleConfigSchema() ModelRoleConfigSchema {
var largeContextFallback *ModelRoleConfigSchema
if m.LargeContextFallback != nil {
@@ -662,7 +689,7 @@ type RoleJSON any
type ClientModelPackSchemaRoles struct {
SchemaUrl SchemaUrl `json:"$schema,omitempty"`
- LocalProvider ModelProvider `json:"local-provider,omitempty"`
+ LocalProvider ModelProvider `json:"localProvider,omitempty"`
// in the JSON, these can either be a role as a string or a ModelRoleConfigSchema object for more complex config
Planner RoleJSON `json:"planner"`
diff --git a/app/shared/ai_models_packs.go b/app/shared/ai_models_packs.go
index 55ec3584..ac5fb684 100644
--- a/app/shared/ai_models_packs.go
+++ b/app/shared/ai_models_packs.go
@@ -12,13 +12,13 @@ var StrongModelOpus ModelPack
var AnthropicModelPack ModelPack
var OpenAIModelPack ModelPack
-var GeminiPreviewModelPack ModelPack
-var GeminiExperimentalModelPack ModelPack
+var GeminiModelPack ModelPack
var R1PlannerModelPack ModelPack
var PerplexityPlannerModelPack ModelPack
var OllamaExperimentalModelPack ModelPack
-var OllamaAdaptiveModelPack ModelPack
+var OllamaAdaptiveOssModelPack ModelPack
+var OllamaAdaptiveCommercialModelPack ModelPack
var BuiltInModelPacks = []*ModelPack{
&DailyDriverModelPack,
@@ -27,13 +27,13 @@ var BuiltInModelPacks = []*ModelPack{
&CheapModelPack,
&OSSModelPack,
&OllamaExperimentalModelPack,
- &OllamaAdaptiveModelPack,
+ &OllamaAdaptiveOssModelPack,
+ &OllamaAdaptiveCommercialModelPack,
&OpusPlannerModelPack,
&StrongModelOpus,
&AnthropicModelPack,
&OpenAIModelPack,
- &GeminiPreviewModelPack,
- &GeminiExperimentalModelPack,
+ &GeminiModelPack,
&R1PlannerModelPack,
&PerplexityPlannerModelPack,
}
@@ -81,19 +81,19 @@ func getStrongModelFallback(role ModelRole, modelId ModelId, fns ...func(*ModelR
}
var (
- DailyDriverSchema ModelPackSchema
- ReasoningSchema ModelPackSchema
- StrongSchema ModelPackSchema
- OssSchema ModelPackSchema
- CheapSchema ModelPackSchema
- OllamaExperimentalSchema ModelPackSchema
- OllamaAdaptiveSchema ModelPackSchema
- AnthropicSchema ModelPackSchema
- OpenAISchema ModelPackSchema
- GeminiPreviewSchema ModelPackSchema
- GeminiExperimentalSchema ModelPackSchema
- R1PlannerSchema ModelPackSchema
- PerplexityPlannerSchema ModelPackSchema
+ DailyDriverSchema ModelPackSchema
+ ReasoningSchema ModelPackSchema
+ StrongSchema ModelPackSchema
+ OssSchema ModelPackSchema
+ CheapSchema ModelPackSchema
+ OllamaExperimentalSchema ModelPackSchema
+ OllamaAdaptiveOssSchema ModelPackSchema
+ OllamaAdaptiveCommercialSchema ModelPackSchema
+ AnthropicSchema ModelPackSchema
+ OpenAISchema ModelPackSchema
+ GeminiSchema ModelPackSchema
+ R1PlannerSchema ModelPackSchema
+ PerplexityPlannerSchema ModelPackSchema
)
var BuiltInModelPackSchemas = []*ModelPackSchema{
@@ -103,11 +103,11 @@ var BuiltInModelPackSchemas = []*ModelPackSchema{
&CheapSchema,
&OssSchema,
&OllamaExperimentalSchema,
- &OllamaAdaptiveSchema,
+ &OllamaAdaptiveOssSchema,
+ &OllamaAdaptiveCommercialSchema,
&AnthropicSchema,
&OpenAISchema,
- &GeminiPreviewSchema,
- &GeminiExperimentalSchema,
+ &GeminiSchema,
&R1PlannerSchema,
&PerplexityPlannerSchema,
}
@@ -122,12 +122,12 @@ func init() {
Description: "A mix of models from Anthropic, OpenAI, and Google that balances speed, quality, and cost. Supports up to 2M context.",
ModelPackSchemaRoles: ModelPackSchemaRoles{
Planner: getModelRoleConfig(ModelRolePlanner, "anthropic/claude-sonnet-4",
- getLargeContextFallback(ModelRolePlanner, "google/gemini-2.5-pro-preview",
+ getLargeContextFallback(ModelRolePlanner, "google/gemini-2.5-pro",
getLargeContextFallback(ModelRolePlanner, "google/gemini-pro-1.5"),
),
),
Architect: Pointer(getModelRoleConfig(ModelRoleArchitect, "anthropic/claude-sonnet-4",
- getLargeContextFallback(ModelRoleArchitect, "google/gemini-2.5-pro-preview",
+ getLargeContextFallback(ModelRoleArchitect, "google/gemini-2.5-pro",
getLargeContextFallback(ModelRoleArchitect, "google/gemini-pro-1.5"),
),
)),
@@ -194,27 +194,27 @@ func init() {
OssSchema = ModelPackSchema{
Name: "oss",
- Description: "An experimental mix of the best open source models for coding. Supports up to 56k context, 8k per file. Works best with smaller projects and files. Includes reasoning.",
+ Description: "An experimental mix of the best open source models for coding. Supports up to 144k context, 33k per file.",
ModelPackSchemaRoles: ModelPackSchemaRoles{
Planner: getModelRoleConfig(ModelRolePlanner, "deepseek/r1"),
- Coder: Pointer(getModelRoleConfig(ModelRoleCoder, "deepseek/v3-0324")),
+ Coder: Pointer(getModelRoleConfig(ModelRoleCoder, "deepseek/v3")),
PlanSummary: getModelRoleConfig(ModelRolePlanSummary, "deepseek/r1-hidden"),
Builder: getModelRoleConfig(ModelRoleBuilder, "deepseek/r1-hidden"),
WholeFileBuilder: Pointer(getModelRoleConfig(ModelRoleWholeFileBuilder,
"deepseek/r1-hidden")),
- Namer: getModelRoleConfig(ModelRoleName, "qwen/qwen-2.5-coder-32b-instruct"),
- CommitMsg: getModelRoleConfig(ModelRoleCommitMsg, "qwen/qwen-2.5-coder-32b-instruct"),
+ Namer: getModelRoleConfig(ModelRoleName, "qwen/qwen3-8b"),
+ CommitMsg: getModelRoleConfig(ModelRoleCommitMsg, "qwen/qwen3-8b"),
ExecStatus: getModelRoleConfig(ModelRoleExecStatus, "deepseek/r1-hidden"),
},
}
OllamaExperimentalSchema = ModelPackSchema{
Name: "ollama-experimental",
- Description: "Ollama experimental local blend. Supports up to 110k context. Uses Qwen3-32b model for heavy lifting, Qwen3-14b for summaries, Qwen3-8b for lighter tasks.",
+ Description: "Ollama experimental local blend. Supports up to 110k context. For now, more for experimentation and benchmarking than getting work done.",
ModelPackSchemaRoles: ModelPackSchemaRoles{
LocalProvider: ModelProviderOllama,
- Planner: getModelRoleConfig(ModelRolePlanner, "mistral/devstral-small"),
- PlanSummary: getModelRoleConfig(ModelRolePlanSummary, "qwen/qwen3-14b"),
+ Planner: getModelRoleConfig(ModelRolePlanner, "qwen/qwen3-32b"),
+ PlanSummary: getModelRoleConfig(ModelRolePlanSummary, "mistral/devstral-small"),
Builder: getModelRoleConfig(ModelRoleBuilder, "mistral/devstral-small"),
WholeFileBuilder: Pointer(getModelRoleConfig(ModelRoleWholeFileBuilder,
"mistral/devstral-small")),
@@ -224,21 +224,23 @@ func init() {
},
}
- OllamaAdaptiveSchema = ModelPackSchema{
- Name: "ollama-adaptive",
- Description: "Ollama adaptive blend. Uses local models for planning and context selection, cloud models for implementation and file edits. Supports up to 110k context.",
- ModelPackSchemaRoles: ModelPackSchemaRoles{
- LocalProvider: ModelProviderOllama,
- Planner: getModelRoleConfig(ModelRolePlanner, "mistral/devstral-small"),
- PlanSummary: getModelRoleConfig(ModelRolePlanSummary, "qwen/qwen3-14b"),
- Builder: getModelRoleConfig(ModelRoleBuilder, "mistral/devstral-small"),
- WholeFileBuilder: Pointer(getModelRoleConfig(ModelRoleWholeFileBuilder,
- "mistral/devstral-small")),
- Namer: getModelRoleConfig(ModelRoleName, "qwen/qwen3-8b"),
- CommitMsg: getModelRoleConfig(ModelRoleCommitMsg, "qwen/qwen3-8b"),
- ExecStatus: getModelRoleConfig(ModelRoleExecStatus, "mistral/devstral-small"),
- },
- }
+ // Copy daily driver schema and modify it to use ollama for lighter tasks
+ OllamaAdaptiveCommercialSchema = cloneSchema(DailyDriverSchema)
+ OllamaAdaptiveCommercialSchema.Name = "ollama-adaptive-commercial"
+ OllamaAdaptiveCommercialSchema.Description = "Ollama adaptive commercial blend. Uses 'daily-driver' for heavy lifting, local models for lighter tasks."
+ OllamaAdaptiveCommercialSchema.LocalProvider = ModelProviderOllama
+ OllamaAdaptiveCommercialSchema.PlanSummary = getModelRoleConfig(ModelRolePlanSummary, "mistral/devstral-small")
+ OllamaAdaptiveCommercialSchema.CommitMsg = getModelRoleConfig(ModelRoleCommitMsg, "qwen/qwen3-8b")
+ OllamaAdaptiveCommercialSchema.Namer = getModelRoleConfig(ModelRoleName, "qwen/qwen3-8b")
+
+ // Copy oss schema and modify it to use ollama for lighter tasks
+ OllamaAdaptiveOssSchema = cloneSchema(OssSchema)
+ OllamaAdaptiveOssSchema.Name = "ollama-adaptive-oss"
+ OllamaAdaptiveOssSchema.Description = "Ollama adaptive OSS blend. Uses local models for planning and context selection, open source cloud models for implementation and file edits. Supports up to 110k context."
+ OllamaAdaptiveOssSchema.LocalProvider = ModelProviderOllama
+ OllamaAdaptiveOssSchema.PlanSummary = getModelRoleConfig(ModelRolePlanSummary, "mistral/devstral-small")
+ OllamaAdaptiveOssSchema.CommitMsg = getModelRoleConfig(ModelRoleCommitMsg, "qwen/qwen3-8b")
+ OllamaAdaptiveOssSchema.Namer = getModelRoleConfig(ModelRoleName, "qwen/qwen3-8b")
OpenAISchema = ModelPackSchema{
Name: "openai",
@@ -271,28 +273,12 @@ func init() {
},
}
- GeminiPreviewSchema = ModelPackSchema{
+ GeminiSchema = ModelPackSchema{
Name: "gemini-preview",
Description: "Uses Gemini 2.5 Pro Preview for planning and coding, default models for other roles. Supports up to 1M input context.",
ModelPackSchemaRoles: ModelPackSchemaRoles{
- Planner: getModelRoleConfig(ModelRolePlanner, "google/gemini-2.5-pro-preview"),
- Coder: Pointer(getModelRoleConfig(ModelRoleCoder, "google/gemini-2.5-pro-preview")),
- PlanSummary: getModelRoleConfig(ModelRolePlanSummary, "openai/o4-mini-low"),
- Builder: defaultBuilder,
- WholeFileBuilder: Pointer(getModelRoleConfig(ModelRoleWholeFileBuilder,
- "openai/o4-mini-medium")),
- Namer: getModelRoleConfig(ModelRoleName, "openai/gpt-4.1-mini"),
- CommitMsg: getModelRoleConfig(ModelRoleCommitMsg, "openai/gpt-4.1-mini"),
- ExecStatus: getModelRoleConfig(ModelRoleExecStatus, "openai/o4-mini-low"),
- },
- }
-
- GeminiExperimentalSchema = ModelPackSchema{
- Name: "gemini-exp",
- Description: "Uses Gemini 2.5 Pro Experimental (free) for planning and coding, default models for other roles. Supports up to 1M input context.",
- ModelPackSchemaRoles: ModelPackSchemaRoles{
- Planner: getModelRoleConfig(ModelRolePlanner, "google/gemini-2.5-pro-exp"),
- Coder: Pointer(getModelRoleConfig(ModelRoleCoder, "google/gemini-2.5-pro-exp")),
+ Planner: getModelRoleConfig(ModelRolePlanner, "google/gemini-2.5-pro"),
+ Coder: Pointer(getModelRoleConfig(ModelRoleCoder, "google/gemini-2.5-pro")),
PlanSummary: getModelRoleConfig(ModelRolePlanSummary, "openai/o4-mini-low"),
Builder: defaultBuilder,
WholeFileBuilder: Pointer(getModelRoleConfig(ModelRoleWholeFileBuilder,
@@ -341,11 +327,11 @@ func init() {
CheapModelPack = CheapSchema.ToModelPack()
OSSModelPack = OssSchema.ToModelPack()
OllamaExperimentalModelPack = OllamaExperimentalSchema.ToModelPack()
- OllamaAdaptiveModelPack = OllamaAdaptiveSchema.ToModelPack()
+ OllamaAdaptiveOssModelPack = OllamaAdaptiveOssSchema.ToModelPack()
+ OllamaAdaptiveCommercialModelPack = OllamaAdaptiveCommercialSchema.ToModelPack()
AnthropicModelPack = AnthropicSchema.ToModelPack()
OpenAIModelPack = OpenAISchema.ToModelPack()
- GeminiPreviewModelPack = GeminiPreviewSchema.ToModelPack()
- GeminiExperimentalModelPack = GeminiExperimentalSchema.ToModelPack()
+ GeminiModelPack = GeminiSchema.ToModelPack()
R1PlannerModelPack = R1PlannerSchema.ToModelPack()
PerplexityPlannerModelPack = PerplexityPlannerSchema.ToModelPack()
@@ -356,11 +342,11 @@ func init() {
&CheapModelPack,
&OSSModelPack,
&OllamaExperimentalModelPack,
- &OllamaAdaptiveModelPack,
+ &OllamaAdaptiveOssModelPack,
+ &OllamaAdaptiveCommercialModelPack,
&AnthropicModelPack,
&OpenAIModelPack,
- &GeminiPreviewModelPack,
- &GeminiExperimentalModelPack,
+ &GeminiModelPack,
&R1PlannerModelPack,
&PerplexityPlannerModelPack,
}
@@ -376,3 +362,24 @@ func init() {
}
}
+
+// pointer fields need to be cloned to avoid modifying the original schema
+func cloneSchema(schema ModelPackSchema) ModelPackSchema {
+ res := schema
+
+ if schema.Architect != nil {
+ tmp := *schema.Architect
+ res.Architect = &tmp
+ }
+
+ if schema.Coder != nil {
+ tmp := *schema.Coder
+ res.Coder = &tmp
+ }
+ if schema.WholeFileBuilder != nil {
+ tmp := *schema.WholeFileBuilder
+ res.WholeFileBuilder = &tmp
+ }
+
+ return res
+}
diff --git a/app/shared/ai_models_providers.go b/app/shared/ai_models_providers.go
index d7d9d29b..1769af77 100644
--- a/app/shared/ai_models_providers.go
+++ b/app/shared/ai_models_providers.go
@@ -295,13 +295,9 @@ func GetProvidersForAuthVarsWithModelId(authVars map[string]string, settings *Pl
continue
}
- if localProvider != "" {
- if !provider.LocalOnly {
- continue
- }
- if provider.Provider != localProvider {
- continue
- }
+ // if the model pack has a local provider, the provider is local only, and the provider is not the local provider, skip it
+ if localProvider != "" && provider.LocalOnly && provider.Provider != localProvider {
+ continue
}
res = append(res, provider)
diff --git a/docs/docs/cli-reference.md b/docs/docs/cli-reference.md
index b93e0684..75989a7b 100644
--- a/docs/docs/cli-reference.md
+++ b/docs/docs/cli-reference.md
@@ -67,7 +67,7 @@ The REPL has a few convenient flags you can use to start it with different modes
--strong Strong pack (more capable models, higher cost and slower)
--cheap Cheap pack (less capable models, lower cost and faster)
--oss Open source pack (open source models)
- --gemini-preview Gemini preview pack (Gemini 2.5 Pro Preview for planning and coding, default models for other roles)
+ --gemini Gemini pack (Gemini 2.5 Pro Preview for planning and coding, default models for other roles)
```
All commands listed below can be run in the REPL by prefixing them with a backslash (`\`), e.g. `\new`.
@@ -107,7 +107,7 @@ plandex new -n new-plan # with name
`--oss`: Start the plan with the open source model pack.
-`--gemini-preview`: Start the plan with the Gemini preview model pack.
+`--gemini`: Start the plan with the Gemini model pack.
### plans
@@ -747,7 +747,7 @@ plandex set-model reasoning
plandex set-model strong
plandex set-model cheap
plandex set-model oss
-plandex set-model gemini-preview
+plandex set-model gemini
```
### set-model default
diff --git a/docs/docs/environment-variables.md b/docs/docs/environment-variables.md
index 2fd07f95..abef8042 100644
--- a/docs/docs/environment-variables.md
+++ b/docs/docs/environment-variables.md
@@ -87,20 +87,14 @@ GOENV=development # Whether to run in development or production mode. Must be 'd
PLANDEX_BASE_DIR= # The base directory to read and write files. Defaults to '$HOME/plandex-server' in development mode, '/plandex-server' in production.
API_HOST= # The host the API server listens on. Defaults to 'http://localhost:$PORT'. In production mode, should be a host like 'https://api.your-domain.ai'.
PORT=8099 # The port the server listens on. Defaults to 8099.
+DATABASE_URL= # The URL of the PostgreSQL database. Defaults to 'postgres://plandex:plandex@plandex-postgres:5432/plandex?sslmode=disable' in development mode
+LOCAL_MODE= # Whether to run in local mode
+OLLAMA_BASE_URL= # The base URL of the Ollama server—only need when the server is running in a Docker container and needs to access Ollama models running outside of the container
```
### docker-compose
-For self-hosting with docker-compose, default environment variables are set in `app/_env`. This file should be copied to `app/.env` before running the server. You can override any of these defaults in `.env`.
-
-```bash
-PLANDEX_DATA_DIR=/var/lib/plandex/data # When using docker-compose, this is the directory *on your machine* that the Plandex server will use to store data—it will be mounted to the Docker container as a volume.
-
-# Database Credentials
-POSTGRES_DATABASE=plandex # Your postgres database.
-POSTGRES_USER=plandex # Your postgres user.
-POSTGRES_PASSWORD=plandex # Your postgres password.
-```
+For self-hosting with docker-compose, default values for all necessary environment variables are set in the `app/docker-compose.yml` file. This file is designed to be used with [local mode](./hosting/self-hosting/local-mode-quickstart.md), but you can adapt it to your needs.
### Other methods
diff --git a/docs/docs/models/custom-models.md b/docs/docs/models/custom-models.md
index 87a810b5..5bf677d9 100644
--- a/docs/docs/models/custom-models.md
+++ b/docs/docs/models/custom-models.md
@@ -97,6 +97,7 @@ Define models with their capabilities and provider mappings:
```json
{
+ "$schema": "https://plandex.ai/schemas/models-input.schema.json",
"models": [
{
"modelId": "my-model",
@@ -138,6 +139,7 @@ Create your own combinations of models for different roles:
```json
{
+ "$schema": "https://plandex.ai/schemas/models-input.schema.json",
"modelPacks": [
{
"name": "custom-pack",
@@ -146,7 +148,18 @@ Create your own combinations of models for different roles:
"coder": "anthropic/claude-sonnet-4",
"architect": "custom-model-publisher/model-name",
"summarizer": "openai/gpt-4.1",
- "builder": "anthropic/claude-sonnet-4",
+ "builder": {
+ "modelId": "anthropic/claude-sonnet-4",
+ "strongModel": "openai/o3-medium"
+ },
+ "wholeFileBuilder": {
+ "modelId": "anthropic/claude-sonnet-4",
+ "largeContextFallback": {
+ "modelId": "google/gemini-2.5-pro",
+ "largeOutputFallback": "openai/o4-mini-low"
+ },
+ "errorFallback": "openai/gpt-4.1"
+ },
"names": "openai/gpt-4.1-mini",
"commitMessages": "openai/gpt-4.1-mini",
"autoContinue": "openai/o4-mini-medium"
diff --git a/docs/docs/models/model-settings.md b/docs/docs/models/model-settings.md
index 9bc7a9f1..b7ded8d6 100644
--- a/docs/docs/models/model-settings.md
+++ b/docs/docs/models/model-settings.md
@@ -37,6 +37,7 @@ Model roles can either be a string (the model ID) or an object with model config
```json
{
+ "$schema": "https://plandex.ai/schemas/model-pack-inline.schema.json",
"planner": "anthropic/claude-opus-4",
"coder": "anthropic/claude-sonnet-4",
"architect": "anthropic/claude-sonnet-4",
@@ -54,11 +55,12 @@ You can also configure individual role settings and fallbacks with an object:
```json
{
+ "$schema": "https://plandex.ai/schemas/model-pack-inline.schema.json",
"planner": {
"modelId": "anthropic/claude-opus-4",
"temperature": 0.7,
"topP": 0.9,
- "largeContextFallback": "google/gemini-2.5-pro-preview"
+ "largeContextFallback": "google/gemini-2.5-pro"
},
"coder": "anthropic/claude-sonnet-4",
"architect": "anthropic/claude-sonnet-4",
@@ -67,6 +69,13 @@ You can also configure individual role settings and fallbacks with an object:
"modelId": "anthropic/claude-sonnet-4",
"errorFallback": "openai/gpt-4.1"
},
+ "wholeFileBuilder": {
+ "modelId": "anthropic/claude-sonnet-4",
+ "largeContextFallback": {
+ "modelId": "google/gemini-2.5-pro",
+ "largeOutputFallback": "openai/o4-mini-low"
+ }
+ },
"names": "anthropic/claude-3.5-haiku",
"commitMessages": "anthropic/claude-3.5-haiku",
"autoContinue": "anthropic/claude-3.5-haiku"
@@ -93,6 +102,7 @@ You can set the top-level `localProvider` key to `ollama` to use local models vi
```json
{
+ "$schema": "https://plandex.ai/schemas/model-pack-inline.schema.json",
"localProvider": "ollama",
"planner": "ollama/deepseek-r1:14b",
...