x models: add glm 4.7 flash to mlx engine

docs/integrations/claude-code.mdx

@@ -2,7 +2,7 @@
 title: Claude Code
 ---
 
-Claude Code is Anthropic's agentic coding tool that can read, modify, and execute code in your working directory.
+Claude Code is Anthropic's agentic coding tool that can read, modify, and execute code in your working directory. 
 
 Open models can be used with Claude Code through Ollama's Anthropic-compatible API, enabling you to use models such as `qwen3-coder`, `gpt-oss:20b`, or other models.
 
@@ -26,16 +26,6 @@ irm https://claude.ai/install.ps1 | iex
 
 ## Usage with Ollama
 
-Configure Claude Code to use Ollama:
-
-```shell
-ollama config claude
-```
-
-This will prompt you to select a model and automatically configure Claude Code to use Ollama.
-
-<Accordion title="Manual Configuration">
-
 Claude Code connects to Ollama using the Anthropic-compatible API.
 
 1. Set the environment variables:
@@ -57,9 +47,7 @@ Or run with environment variables inline:
 ANTHROPIC_AUTH_TOKEN=ollama ANTHROPIC_BASE_URL=http://localhost:11434 claude --model gpt-oss:20b
 ```
 
-</Accordion>
-
-<Note>Claude Code requires a large context window. We recommend at least 32K tokens. See the [context length documentation](/context-length) for how to adjust context length in Ollama.</Note>
+**Note:** Claude Code requires a large context window. We recommend at least 32K tokens. See the [context length documentation](/context-length) for how to adjust context length in Ollama.
 
 ## Connecting to ollama.com
 
@@ -87,4 +75,4 @@ claude --model glm-4.7:cloud
 ### Local models
 - `qwen3-coder` - Excellent for coding tasks
 - `gpt-oss:20b` - Strong general-purpose model
-- `gpt-oss:120b` - Larger general-purpose model for more complex tasks
+- `gpt-oss:120b` - Larger general-purpose model for more complex tasks

docs/integrations/codex.mdx

@@ -2,31 +2,22 @@
 title: Codex
 ---
 
-Codex is OpenAI's agentic coding tool for the command line.
 
 ## Install
 
 Install the [Codex CLI](https://developers.openai.com/codex/cli/):
 
-```shell
+```
 npm install -g @openai/codex
 ```
 
 ## Usage with Ollama
 
-Configure Codex to use Ollama:
-
-```shell
-ollama config codex
-```
-
-This will prompt you to select a model and automatically configure Codex to use Ollama.
-
-<Accordion title="Manual Configuration">
+<Note>Codex requires a larger context window. It is recommended to use a context window of at least 32K tokens.</Note>
 
 To use `codex` with Ollama, use the `--oss` flag:
 
-```shell
+```
 codex --oss
 ```
 
@@ -34,22 +25,20 @@ codex --oss
 
 By default, codex will use the local `gpt-oss:20b` model. However, you can specify a different model with the `-m` flag:
 
-```shell
+```
 codex --oss -m gpt-oss:120b
 ```
 
 ### Cloud Models
 
-```shell
+```
 codex --oss -m gpt-oss:120b-cloud
 ```
 
-</Accordion>
-
-<Note>Codex requires a larger context window. It is recommended to use a context window of at least 32K tokens.</Note>
 
 ## Connecting to ollama.com
 
+
 Create an [API key](https://ollama.com/settings/keys) from ollama.com and export it as `OLLAMA_API_KEY`.
 
 To use ollama.com directly, edit your `~/.codex/config.toml` file to point to ollama.com.

docs/integrations/droid.mdx

@@ -2,7 +2,6 @@
 title: Droid
 ---
 
-Droid is Factory's agentic coding tool for the command line.
 
 ## Install
 
@@ -12,80 +11,66 @@ Install the [Droid CLI](https://factory.ai/):
 curl -fsSL https://app.factory.ai/cli | sh
 ```
 
+<Note>Droid requires a larger context window. It is recommended to use a context window of at least 32K tokens. See [Context length](/context-length) for more information.</Note>
+
 ## Usage with Ollama
 
-Configure Droid to use Ollama:
-
-```shell
-ollama config droid
-```
-
-This will prompt you to select models and automatically configure Droid to use Ollama.
-
-<Accordion title="Manual Configuration">
-
-Add a local configuration block to `~/.factory/settings.json`:
+Add a local configuration block to `~/.factory/config.json`:
 
 ```json
 {
-  "customModels": [
+  "custom_models": [
     {
+      "model_display_name": "qwen3-coder [Ollama]",
       "model": "qwen3-coder",
-      "displayName": "qwen3-coder [Ollama]",
-      "baseUrl": "http://localhost:11434/v1",
-      "apiKey": "ollama",
+      "base_url": "http://localhost:11434/v1/",
+      "api_key": "not-needed",
       "provider": "generic-chat-completion-api",
-      "maxOutputTokens": 32000
+      "max_tokens": 32000 
     }
   ]
 }
 ```
 
-Adjust `maxOutputTokens` based on your model's context length (the automated setup detects this automatically).
-
-### Cloud Models
 
+## Cloud Models
 `qwen3-coder:480b-cloud` is the recommended model for use with Droid.
 
-Add the cloud configuration block to `~/.factory/settings.json`:
+Add the cloud configuration block to `~/.factory/config.json`:
 
 ```json
 {
-  "customModels": [
+  "custom_models": [
     {
+      "model_display_name": "qwen3-coder [Ollama Cloud]",
       "model": "qwen3-coder:480b-cloud",
-      "displayName": "qwen3-coder:480b-cloud [Ollama]",
-      "baseUrl": "http://localhost:11434/v1",
-      "apiKey": "ollama",
+      "base_url": "http://localhost:11434/v1/",
+      "api_key": "not-needed",
       "provider": "generic-chat-completion-api",
-      "maxOutputTokens": 128000
+      "max_tokens": 128000
     }
   ]
 }
 ```
 
-</Accordion>
-
-<Note>Droid requires a larger context window. It is recommended to use a context window of at least 32K tokens. See [Context length](/context-length) for more information.</Note>
-
 ## Connecting to ollama.com
 
 1. Create an [API key](https://ollama.com/settings/keys) from ollama.com and export it as `OLLAMA_API_KEY`.
-2. Add the cloud configuration block to `~/.factory/settings.json`:
+2. Add the cloud configuration block to `~/.factory/config.json`:
 
    ```json
    {
-     "customModels": [
+     "custom_models": [
        {
+         "model_display_name": "qwen3-coder [Ollama Cloud]",
          "model": "qwen3-coder:480b",
-         "displayName": "qwen3-coder:480b [Ollama Cloud]",
-         "baseUrl": "https://ollama.com/v1",
-         "apiKey": "OLLAMA_API_KEY",
+         "base_url": "https://ollama.com/v1/",
+         "api_key": "OLLAMA_API_KEY",
          "provider": "generic-chat-completion-api",
-         "maxOutputTokens": 128000
+         "max_tokens": 128000
        }
      ]
    }
    ```
 
-Run `droid` in a new terminal to load the new settings.
+Run `droid` in a new terminal to load the new settings.

docs/integrations/opencode.mdx

@@ -1,63 +0,0 @@
----
-title: OpenCode
----
-
-OpenCode is an agentic coding tool for the terminal.
-
-## Install
-
-Install [OpenCode](https://opencode.ai):
-
-```shell
-curl -fsSL https://opencode.ai/install | bash
-```
-
-## Usage with Ollama
-
-Configure OpenCode to use Ollama:
-
-```shell
-ollama config opencode
-```
-
-This will prompt you to select models and automatically configure OpenCode to use Ollama.
-
-<Accordion title="Manual Configuration">
-
-Add the Ollama provider to `~/.config/opencode/opencode.json`:
-
-```json
-{
-  "$schema": "https://opencode.ai/config.json",
-  "provider": {
-    "ollama": {
-      "npm": "@ai-sdk/openai-compatible",
-      "name": "Ollama (local)",
-      "options": {
-        "baseURL": "http://localhost:11434/v1"
-      },
-      "models": {
-        "qwen3-coder": {
-          "name": "qwen3-coder [Ollama]"
-        }
-      }
-    }
-  }
-}
-```
-
-</Accordion>
-
-<Note>OpenCode requires a larger context window. It is recommended to use a context window of at least 32K tokens. See [Context length](/context-length) for more information.</Note>
-
-## Recommended Models
-
-### Cloud models
-- `qwen3-coder:480b` - Large coding model
-- `glm-4.7:cloud` - High-performance cloud model
-- `minimax-m2.1:cloud` - Fast cloud model
-
-### Local models
-- `qwen3-coder` - Excellent for coding tasks
-- `gpt-oss:20b` - Strong general-purpose model
-- `gpt-oss:120b` - Larger general-purpose model for more complex tasks

x/imagegen/cmd/engine/main.go

@@ -19,6 +19,7 @@ import (
 	"github.com/ollama/ollama/x/imagegen/mlx"
 	"github.com/ollama/ollama/x/imagegen/models/flux2"
 	"github.com/ollama/ollama/x/imagegen/models/gemma3"
+	"github.com/ollama/ollama/x/imagegen/models/glm4_moe_lite"
 	"github.com/ollama/ollama/x/imagegen/models/gpt_oss"
 	"github.com/ollama/ollama/x/imagegen/models/llama"
 	"github.com/ollama/ollama/x/imagegen/models/zimage"
@@ -242,6 +243,8 @@ func load(modelPath string) (Model, error) {
 		return gemma3.Load(modelPath)
 	case "gemma3_text":
 		return gemma3.LoadText(modelPath)
+	case "glm4_moe_lite":
+		return glm4_moe_lite.Load(modelPath)
 	default:
 		return llama.Load(modelPath)
 	}

x/imagegen/mlx/mlx.go

@@ -991,6 +991,19 @@ func Concat(a, b *Array, axis int) *Array {
 	return Concatenate([]*Array{a, b}, axis)
 }
 
+// Stack stacks arrays along a new axis (axis 0 by default)
+func Stack(arrays []*Array, axis int) *Array {
+	handles := make([]C.mlx_array, len(arrays))
+	for i, arr := range arrays {
+		handles[i] = arr.c
+	}
+	vec := C.mlx_vector_array_new_data(&handles[0], C.size_t(len(handles)))
+	res := C.mlx_array_new()
+	C.mlx_stack_axis(&res, vec, C.int(axis), C.default_stream())
+	C.mlx_vector_array_free(vec)
+	return newArray(res)
+}
+
 // Slice slices the array
 func Slice(a *Array, start, stop []int32) *Array {
 	n := len(start)

x/imagegen/models/glm4_moe_lite/glm4_moe_lite.go

@@ -0,0 +1,529 @@
+//go:build mlx
+
+// Package glm4_moe_lite provides the GLM4-MoE-Lite implementation for MLX.
+// This model uses Multi-head Latent Attention (MLA) and Mixture of Experts (MoE).
+package glm4_moe_lite
+
+import (
+	"encoding/json"
+	"fmt"
+	"math"
+	"os"
+	"path/filepath"
+
+	"github.com/ollama/ollama/x/imagegen/cache"
+	"github.com/ollama/ollama/x/imagegen/mlx"
+	"github.com/ollama/ollama/x/imagegen/nn"
+	"github.com/ollama/ollama/x/imagegen/safetensors"
+	"github.com/ollama/ollama/x/imagegen/tokenizer"
+)
+
+// Config holds GLM4-MoE-Lite model configuration
+type Config struct {
+	HiddenSize            int32   `json:"hidden_size"`
+	NumHiddenLayers       int32   `json:"num_hidden_layers"`
+	IntermediateSize      int32   `json:"intermediate_size"`
+	MoEIntermediateSize   int32   `json:"moe_intermediate_size"`
+	NumAttentionHeads     int32   `json:"num_attention_heads"`
+	NumKeyValueHeads      int32   `json:"num_key_value_heads"`
+	VocabSize             int32   `json:"vocab_size"`
+	RMSNormEps            float32 `json:"rms_norm_eps"`
+	RopeTheta             float32 `json:"rope_theta"`
+	MaxPositionEmbeddings int32   `json:"max_position_embeddings"`
+	AttentionBias         bool    `json:"attention_bias"`
+
+	// MLA (Multi-head Latent Attention) parameters
+	QLoraRank      int32 `json:"q_lora_rank"`
+	KVLoraRank     int32 `json:"kv_lora_rank"`
+	QKRopeHeadDim  int32 `json:"qk_rope_head_dim"`
+	QKNopeHeadDim  int32 `json:"qk_nope_head_dim"`
+	VHeadDim       int32 `json:"v_head_dim"`
+
+	// MoE parameters
+	NRoutedExperts      int32   `json:"n_routed_experts"`
+	NSharedExperts      int32   `json:"n_shared_experts"`
+	NumExpertsPerTok    int32   `json:"num_experts_per_tok"`
+	RoutedScalingFactor float32 `json:"routed_scaling_factor"`
+	NormTopKProb        bool    `json:"norm_topk_prob"`
+	FirstKDenseReplace  int32   `json:"first_k_dense_replace"`
+	NGroup              int32   `json:"n_group"`
+	TopKGroup           int32   `json:"topk_group"`
+
+	// Computed fields
+	QHeadDim int32   `json:"-"` // qk_nope_head_dim + qk_rope_head_dim
+	Scale    float32 `json:"-"` // 1/sqrt(QHeadDim)
+}
+
+// MLAAttention implements Multi-head Latent Attention
+type MLAAttention struct {
+	// Low-rank query projections
+	QAProj      *nn.Linear  `weight:"self_attn.q_a_proj"`
+	QALayerNorm *nn.RMSNorm `weight:"self_attn.q_a_layernorm"`
+	QBProj      *nn.Linear  `weight:"self_attn.q_b_proj"`
+
+	// Low-rank KV projections (with shared rope component)
+	KVAProjWithMQA *nn.Linear  `weight:"self_attn.kv_a_proj_with_mqa"`
+	KVALayerNorm   *nn.RMSNorm `weight:"self_attn.kv_a_layernorm"`
+	KVBProj        *nn.Linear  `weight:"self_attn.kv_b_proj"`
+
+	// Output projection
+	OProj *nn.Linear `weight:"self_attn.o_proj"`
+}
+
+// Forward computes MLA attention output
+func (a *MLAAttention) Forward(x *mlx.Array, c cache.Cache, B, L int32, cfg *Config) *mlx.Array {
+	// Query path: q_a_proj -> layernorm -> q_b_proj
+	q := a.QAProj.Forward(x)
+	q = a.QALayerNorm.Forward(q, cfg.RMSNormEps)
+	q = a.QBProj.Forward(q)
+
+	// Reshape Q: [B, L, num_heads * q_head_dim] -> [B, num_heads, L, q_head_dim]
+	q = mlx.Reshape(q, B, L, cfg.NumAttentionHeads, cfg.QHeadDim)
+	q = mlx.Transpose(q, 0, 2, 1, 3)
+
+	// Split Q into nope and rope parts
+	qNope := mlx.Slice(q, []int32{0, 0, 0, 0}, []int32{B, cfg.NumAttentionHeads, L, cfg.QKNopeHeadDim})
+	qPE := mlx.Slice(q, []int32{0, 0, 0, cfg.QKNopeHeadDim}, []int32{B, cfg.NumAttentionHeads, L, cfg.QHeadDim})
+
+	// KV path: kv_a_proj_with_mqa -> split -> layernorm -> kv_b_proj
+	compressedKV := a.KVAProjWithMQA.Forward(x)
+
+	// Split into compressed_kv and k_pe (shared rope component)
+	kvCompressed := mlx.Slice(compressedKV, []int32{0, 0, 0}, []int32{B, L, cfg.KVLoraRank})
+	kPE := mlx.Slice(compressedKV, []int32{0, 0, cfg.KVLoraRank}, []int32{B, L, cfg.KVLoraRank + cfg.QKRopeHeadDim})
+
+	// k_pe is shared across heads (MQA-style): [B, L, rope_dim] -> [B, 1, L, rope_dim]
+	kPE = mlx.Reshape(kPE, B, L, 1, cfg.QKRopeHeadDim)
+	kPE = mlx.Transpose(kPE, 0, 2, 1, 3)
+
+	// Apply layernorm and project KV
+	kvCompressed = a.KVALayerNorm.Forward(kvCompressed, cfg.RMSNormEps)
+	kv := a.KVBProj.Forward(kvCompressed)
+
+	// Reshape KV: [B, L, num_heads * (qk_nope_head_dim + v_head_dim)]
+	kv = mlx.Reshape(kv, B, L, cfg.NumAttentionHeads, cfg.QKNopeHeadDim+cfg.VHeadDim)
+	kv = mlx.Transpose(kv, 0, 2, 1, 3)
+
+	// Split into k_nope and values
+	kNope := mlx.Slice(kv, []int32{0, 0, 0, 0}, []int32{B, cfg.NumAttentionHeads, L, cfg.QKNopeHeadDim})
+	values := mlx.Slice(kv, []int32{0, 0, 0, cfg.QKNopeHeadDim}, []int32{B, cfg.NumAttentionHeads, L, cfg.QKNopeHeadDim + cfg.VHeadDim})
+
+	// Apply RoPE to the rope parts only
+	offset := 0
+	if c != nil {
+		offset = c.Offset()
+	}
+	qPE = mlx.RoPE(qPE, int(cfg.QKRopeHeadDim), true, cfg.RopeTheta, 1.0, offset)
+	kPE = mlx.RoPE(kPE, int(cfg.QKRopeHeadDim), true, cfg.RopeTheta, 1.0, offset)
+
+	// Repeat k_pe across all heads
+	kPE = mlx.Tile(kPE, []int32{1, cfg.NumAttentionHeads, 1, 1})
+
+	// Concatenate nope and rope parts
+	queries := mlx.Concatenate([]*mlx.Array{qNope, qPE}, 3)
+	keys := mlx.Concatenate([]*mlx.Array{kNope, kPE}, 3)
+
+	// Update KV cache
+	if c != nil {
+		keys, values = c.Update(keys, values, int(L))
+	}
+
+	// Scaled dot product attention
+	out := mlx.ScaledDotProductAttention(queries, keys, values, cfg.Scale, L > 1)
+
+	// Reshape back: [B, num_heads, L, v_head_dim] -> [B, L, num_heads * v_head_dim]
+	out = mlx.Reshape(mlx.Transpose(out, 0, 2, 1, 3), B, L, cfg.NumAttentionHeads*cfg.VHeadDim)
+
+	return a.OProj.Forward(out)
+}
+
+// DenseMLP implements the standard SwiGLU MLP for dense layers
+type DenseMLP struct {
+	GateProj *nn.Linear `weight:"mlp.gate_proj"`
+	UpProj   *nn.Linear `weight:"mlp.up_proj"`
+	DownProj *nn.Linear `weight:"mlp.down_proj"`
+}
+
+// Forward applies the SwiGLU MLP
+func (m *DenseMLP) Forward(x *mlx.Array) *mlx.Array {
+	gate := mlx.SiLU(m.GateProj.Forward(x))
+	up := m.UpProj.Forward(x)
+	return m.DownProj.Forward(mlx.Mul(gate, up))
+}
+
+// MoEGate implements the expert gating mechanism
+type MoEGate struct {
+	Weight                 *mlx.Array `weight:"mlp.gate.weight"`
+	EScoreCorrectionBias   *mlx.Array `weight:"mlp.gate.e_score_correction_bias,optional"`
+}
+
+// Forward computes expert selection indices and scores
+func (g *MoEGate) Forward(x *mlx.Array, cfg *Config) (*mlx.Array, *mlx.Array) {
+	// Compute gate logits: x @ weight.T
+	gates := mlx.Linear(x, mlx.Transpose(g.Weight, 1, 0))
+
+	// Sigmoid scoring
+	scores := mlx.Sigmoid(gates)
+	origScores := scores
+
+	// Add correction bias if present
+	if g.EScoreCorrectionBias != nil {
+		scores = mlx.Add(scores, g.EScoreCorrectionBias)
+	}
+
+	// Group-wise expert selection (simplified for n_group=1)
+	// Select top-k experts
+	topK := cfg.NumExpertsPerTok
+	negScores := mlx.Neg(scores)
+	inds := mlx.Argpartition(negScores, int(topK)-1, -1)
+
+	shape := inds.Shape()
+	inds = mlx.Slice(inds, []int32{0, 0, 0}, []int32{shape[0], shape[1], topK})
+
+	// Get scores for selected experts
+	scores = mlx.TakeAlongAxis(origScores, inds, -1)
+
+	// Normalize if configured
+	if topK > 1 && cfg.NormTopKProb {
+		sumScores := mlx.Sum(scores, -1, true)
+		scores = mlx.Div(scores, sumScores)
+	}
+
+	// Apply routing scaling factor
+	scores = mlx.MulScalar(scores, cfg.RoutedScalingFactor)
+
+	return inds, scores
+}
+
+// SwitchMLP implements the MoE expert computation using stacked weights
+// Note: No weight tags - these are populated manually by stacking expert weights
+type SwitchMLP struct {
+	GateWeight *mlx.Array
+	UpWeight   *mlx.Array
+	DownWeight *mlx.Array
+}
+
+// Forward applies the switched expert MLP
+func (s *SwitchMLP) Forward(x *mlx.Array, indices *mlx.Array, cfg *Config) *mlx.Array {
+	shape := x.Shape()
+	B, L := shape[0], shape[1]
+	topK := cfg.NumExpertsPerTok
+
+	// Expand x for expert computation: [B, L, D] -> [B, L, 1, 1, D]
+	xExpanded := mlx.ExpandDims(mlx.ExpandDims(x, -2), -2)
+
+	// Flatten for gather_mm: [B*L, 1, 1, D]
+	xFlat := mlx.Reshape(xExpanded, B*L, 1, 1, cfg.HiddenSize)
+
+	// Flatten indices: [B, L, topK] -> [B*L, topK]
+	idxFlat := mlx.Reshape(indices, B*L, topK)
+
+	// Sort for efficient gather (when we have many tokens)
+	doSort := B*L >= 64
+	var invOrder *mlx.Array
+	n := B * L * topK
+
+	if doSort {
+		idxAll := mlx.Flatten(idxFlat)
+		order := mlx.Argsort(idxAll, 0)
+		invOrder = mlx.Argsort(order, 0)
+		// Reorder x based on sorted indices
+		xFlat = mlx.ExpandDims(mlx.Take(mlx.Squeeze(xFlat, 1), mlx.FloorDivideScalar(order, topK), 0), 1)
+		idxFlat = mlx.Reshape(mlx.Take(idxAll, order, 0), n, 1)
+	}
+
+	// Expert computation using gather_mm
+	// gate: x @ gate_weight.T (indices are on the rhs/weight side)
+	gate := mlx.GatherMM(xFlat, mlx.Transpose(s.GateWeight, 0, 2, 1), nil, idxFlat, doSort)
+	// up: x @ up_weight.T
+	up := mlx.GatherMM(xFlat, mlx.Transpose(s.UpWeight, 0, 2, 1), nil, idxFlat, doSort)
+
+	// SwiGLU activation
+	hidden := mlx.Mul(mlx.SiLU(gate), up)
+
+	// down: hidden @ down_weight.T
+	down := mlx.GatherMM(hidden, mlx.Transpose(s.DownWeight, 0, 2, 1), nil, idxFlat, doSort)
+
+	// Unsort if we sorted
+	if doSort {
+		down = mlx.Reshape(mlx.Take(mlx.Squeeze(mlx.Squeeze(down, 2), 1), invOrder, 0), B*L, topK, cfg.HiddenSize)
+	} else {
+		down = mlx.Squeeze(down, 2)
+	}
+
+	return mlx.Reshape(down, B, L, topK, cfg.HiddenSize)
+}
+
+// SharedExperts implements the shared expert MLP
+type SharedExperts struct {
+	GateProj *nn.Linear `weight:"mlp.shared_experts.gate_proj"`
+	UpProj   *nn.Linear `weight:"mlp.shared_experts.up_proj"`
+	DownProj *nn.Linear `weight:"mlp.shared_experts.down_proj"`
+}
+
+// Forward applies the shared expert MLP
+func (s *SharedExperts) Forward(x *mlx.Array) *mlx.Array {
+	gate := mlx.SiLU(s.GateProj.Forward(x))
+	up := s.UpProj.Forward(x)
+	return s.DownProj.Forward(mlx.Mul(gate, up))
+}
+
+// MoE implements the full Mixture of Experts layer
+type MoE struct {
+	Gate          *MoEGate
+	SwitchMLP     *SwitchMLP
+	SharedExperts *SharedExperts
+}
+
+// Forward applies the MoE layer
+func (m *MoE) Forward(x *mlx.Array, cfg *Config) *mlx.Array {
+	shape := x.Shape()
+	B, L := shape[0], shape[1]
+
+	// Get expert indices and scores
+	inds, scores := m.Gate.Forward(x, cfg)
+
+	// Apply routed experts
+	expertOut := m.SwitchMLP.Forward(x, inds, cfg)
+
+	// Weight by scores: [B, L, topK, D] * [B, L, topK, 1] -> sum over topK
+	scoresExpanded := mlx.ExpandDims(scores, -1)
+	y := mlx.Sum(mlx.Mul(expertOut, scoresExpanded), 2, false)
+
+	// Add shared experts if present
+	if m.SharedExperts != nil {
+		y = mlx.Add(y, m.SharedExperts.Forward(x))
+	}
+
+	return mlx.Reshape(y, B, L, cfg.HiddenSize)
+}
+
+// DenseBlock represents a dense transformer block (for first_k_dense_replace layers)
+type DenseBlock struct {
+	Attention              *MLAAttention
+	MLP                    *DenseMLP
+	InputLayerNorm         *nn.RMSNorm `weight:"input_layernorm"`
+	PostAttentionLayerNorm *nn.RMSNorm `weight:"post_attention_layernorm"`
+}
+
+// Forward applies the dense block
+func (b *DenseBlock) Forward(x *mlx.Array, c cache.Cache, B, L int32, cfg *Config) *mlx.Array {
+	// Pre-norm attention with residual
+	r := b.Attention.Forward(b.InputLayerNorm.Forward(x, cfg.RMSNormEps), c, B, L, cfg)
+	h := mlx.Add(x, r)
+
+	// Pre-norm MLP with residual
+	r = b.MLP.Forward(b.PostAttentionLayerNorm.Forward(h, cfg.RMSNormEps))
+	return mlx.Add(h, r)
+}
+
+// MoEBlock represents a MoE transformer block
+type MoEBlock struct {
+	Attention              *MLAAttention
+	MoE                    *MoE
+	InputLayerNorm         *nn.RMSNorm `weight:"input_layernorm"`
+	PostAttentionLayerNorm *nn.RMSNorm `weight:"post_attention_layernorm"`
+}
+
+// Forward applies the MoE block
+func (b *MoEBlock) Forward(x *mlx.Array, c cache.Cache, B, L int32, cfg *Config) *mlx.Array {
+	// Pre-norm attention with residual
+	r := b.Attention.Forward(b.InputLayerNorm.Forward(x, cfg.RMSNormEps), c, B, L, cfg)
+	h := mlx.Add(x, r)
+
+	// Pre-norm MoE with residual
+	r = b.MoE.Forward(b.PostAttentionLayerNorm.Forward(h, cfg.RMSNormEps), cfg)
+	return mlx.Add(h, r)
+}
+
+// Block interface for both dense and MoE blocks
+type Block interface {
+	Forward(x *mlx.Array, c cache.Cache, B, L int32, cfg *Config) *mlx.Array
+}
+
+// Model represents the complete GLM4-MoE-Lite model
+type Model struct {
+	EmbedTokens *nn.Embedding `weight:"model.embed_tokens"`
+	Layers      []Block       `weight:"-"` // Loaded manually due to different block types
+	Norm        *nn.RMSNorm   `weight:"model.norm"`
+	LMHead      *nn.Linear    `weight:"lm_head"`
+
+	tok *tokenizer.Tokenizer
+	*Config
+}
+
+// sanitizeExpertWeights stacks individual expert weights into a single tensor
+func sanitizeExpertWeights(weights *safetensors.ModelWeights, prefix string, numExperts int32) (*mlx.Array, *mlx.Array, *mlx.Array) {
+	var gateWeights, upWeights, downWeights []*mlx.Array
+
+	for e := int32(0); e < numExperts; e++ {
+		gw, _ := weights.GetTensor(fmt.Sprintf("%s.mlp.experts.%d.gate_proj.weight", prefix, e))
+		uw, _ := weights.GetTensor(fmt.Sprintf("%s.mlp.experts.%d.up_proj.weight", prefix, e))
+		dw, _ := weights.GetTensor(fmt.Sprintf("%s.mlp.experts.%d.down_proj.weight", prefix, e))
+
+		if gw != nil {
+			gateWeights = append(gateWeights, gw)
+		}
+		if uw != nil {
+			upWeights = append(upWeights, uw)
+		}
+		if dw != nil {
+			downWeights = append(downWeights, dw)
+		}
+	}
+
+	var stackedGate, stackedUp, stackedDown *mlx.Array
+	if len(gateWeights) > 0 {
+		stackedGate = mlx.Stack(gateWeights, 0)
+	}
+	if len(upWeights) > 0 {
+		stackedUp = mlx.Stack(upWeights, 0)
+	}
+	if len(downWeights) > 0 {
+		stackedDown = mlx.Stack(downWeights, 0)
+	}
+
+	return stackedGate, stackedUp, stackedDown
+}
+
+// Load loads a GLM4-MoE-Lite model from the given path
+func Load(modelPath string) (*Model, error) {
+	data, err := os.ReadFile(filepath.Join(modelPath, "config.json"))
+	if err != nil {
+		return nil, fmt.Errorf("load config: %w", err)
+	}
+
+	var cfg Config
+	if err := json.Unmarshal(data, &cfg); err != nil {
+		return nil, fmt.Errorf("parse config: %w", err)
+	}
+
+	// Compute derived fields
+	cfg.QHeadDim = cfg.QKNopeHeadDim + cfg.QKRopeHeadDim
+	cfg.Scale = float32(1.0 / math.Sqrt(float64(cfg.QHeadDim)))
+
+	weights, err := safetensors.LoadModelWeights(modelPath)
+	if err != nil {
+		return nil, fmt.Errorf("load weights: %w", err)
+	}
+
+	tok, err := tokenizer.Load(filepath.Join(modelPath, "tokenizer.json"))
+	if err != nil {
+		return nil, fmt.Errorf("load tokenizer: %w", err)
+	}
+
+	m := &Model{
+		Layers: make([]Block, cfg.NumHiddenLayers),
+		Config: &cfg,
+		tok:    tok,
+	}
+
+	// Load embedding, norm, and lm_head
+	if err := safetensors.LoadModule(m, weights, ""); err != nil {
+		return nil, err
+	}
+
+	// Load layers manually due to different block types
+	for i := int32(0); i < cfg.NumHiddenLayers; i++ {
+		prefix := fmt.Sprintf("model.layers.%d", i)
+
+		// Load attention (same for both block types)
+		attn := &MLAAttention{}
+		if err := safetensors.LoadModule(attn, weights, prefix); err != nil {
+			return nil, fmt.Errorf("layer %d attention: %w", i, err)
+		}
+
+		if i < cfg.FirstKDenseReplace {
+			// Dense block
+			block := &DenseBlock{Attention: attn}
+			if err := safetensors.LoadModule(block, weights, prefix); err != nil {
+				return nil, fmt.Errorf("layer %d dense: %w", i, err)
+			}
+			m.Layers[i] = block
+		} else {
+			// MoE block
+			block := &MoEBlock{Attention: attn}
+			if err := safetensors.LoadModule(block, weights, prefix); err != nil {
+				return nil, fmt.Errorf("layer %d moe block: %w", i, err)
+			}
+
+			// Stack expert weights
+			gateW, upW, downW := sanitizeExpertWeights(weights, prefix, cfg.NRoutedExperts)
+
+			block.MoE = &MoE{
+				Gate: &MoEGate{},
+				SwitchMLP: &SwitchMLP{
+					GateWeight: gateW,
+					UpWeight:   upW,
+					DownWeight: downW,
+				},
+			}
+
+			// Load gate weights
+			if err := safetensors.LoadModule(block.MoE.Gate, weights, prefix); err != nil {
+				return nil, fmt.Errorf("layer %d gate: %w", i, err)
+			}
+
+			// Load shared experts if present
+			if cfg.NSharedExperts > 0 {
+				block.MoE.SharedExperts = &SharedExperts{}
+				if err := safetensors.LoadModule(block.MoE.SharedExperts, weights, prefix); err != nil {
+					return nil, fmt.Errorf("layer %d shared experts: %w", i, err)
+				}
+			}
+
+			m.Layers[i] = block
+		}
+	}
+
+	mlx.Eval(mlx.Collect(m)...)
+	weights.ReleaseAll()
+
+	return m, nil
+}
+
+// Forward computes the forward pass of the model
+func (m *Model) Forward(tokens *mlx.Array, caches []cache.Cache) *mlx.Array {
+	B, L := tokens.Shape()[0], tokens.Shape()[1]
+
+	h := m.EmbedTokens.Forward(tokens)
+
+	for i, layer := range m.Layers {
+		var c cache.Cache
+		if caches != nil {
+			c = caches[i]
+		}
+		h = layer.Forward(h, c, B, L, m.Config)
+	}
+
+	h = m.Norm.Forward(h, m.RMSNormEps)
+	return m.LMHead.Forward(h)
+}
+
+// Interface methods
+
+// NumLayers returns the number of transformer layers
+func (m *Model) NumLayers() int { return len(m.Layers) }
+
+// MaxContextLength returns the maximum context length
+func (m *Model) MaxContextLength() int32 { return m.MaxPositionEmbeddings }
+
+// VocabSize returns the vocabulary size
+func (m *Model) VocabSize() int32 { return m.Config.VocabSize }
+
+// Tokenizer returns the model's tokenizer
+func (m *Model) Tokenizer() *tokenizer.Tokenizer { return m.tok }
+
+// NewCache creates a new KV cache for the model
+func (m *Model) NewCache(maxSeqLen int32) []cache.Cache {
+	caches := make([]cache.Cache, len(m.Layers))
+	for i := range caches {
+		caches[i] = cache.NewKVCache()
+	}
+	return caches
+}
+
+// FormatPrompt applies the GLM-4 chat template
+func (m *Model) FormatPrompt(prompt string) string {
+	return "[gMASK]<sop><|user|>\n" + prompt + "<|assistant|>\n"
+}