examples: add basic Anthropic API demo

Demonstrates basic Anthropic Claude API usage:
- Simple text completion
- Complex reasoning tasks
- System messages for context setting
- Token usage tracking

Author: tmc

docs/docs/how-to/migration-guides.md

@@ -0,0 +1,206 @@
+# Migration Guides
+
+This document provides migration guides for major changes in LangChainGo.
+
+## OpenAI Default Model Change (v0.1.14+)
+
+### What Changed
+
+Starting with LangChainGo v0.1.14+, the default OpenAI model has changed from **`gpt-3.5-turbo`** to **`gpt-4o-mini`**.
+
+### Why This Change
+
+- `gpt-4o-mini` offers better performance and capabilities than `gpt-3.5-turbo`
+- It's more cost-effective and faster than full GPT-4 models
+- Provides better reasoning and instruction-following capabilities
+
+### Impact Assessment
+
+**If you're not explicitly setting a model:**
+```go
+// This will now use gpt-4o-mini instead of gpt-3.5-turbo
+llm, err := openai.New()
+```
+
+**Your code will continue to work**, but responses might be:
+- Slightly different in style or format
+- More accurate and detailed
+- Potentially use slightly more tokens
+
+### Migration Options
+
+#### Option 1: Accept the new default (Recommended)
+Most applications will benefit from the improved capabilities. No code changes needed.
+
+#### Option 2: Explicitly use the old model
+If you need exact compatibility with previous behavior:
+
+```go
+// Before (implicit)
+llm, err := openai.New()
+
+// After (explicit)
+llm, err := openai.New(openai.WithModel("gpt-3.5-turbo"))
+```
+
+#### Option 3: Upgrade to a better model
+Take this opportunity to upgrade to an even better model:
+
+```go
+// Upgrade to full GPT-4o for best quality
+llm, err := openai.New(openai.WithModel("gpt-4o"))
+
+// Or use GPT-4o-mini explicitly
+llm, err := openai.New(openai.WithModel("gpt-4o-mini"))
+```
+
+### Testing Your Application
+
+1. **Test with default model**: Run your application without specifying a model
+2. **Compare outputs**: Check if the new outputs meet your requirements
+3. **Monitor token usage**: `gpt-4o-mini` usage should be similar to `gpt-3.5-turbo`
+4. **Update tests**: If you have tests that depend on specific response formats
+
+### Example Migration
+
+```go
+package main
+
+import (
+    "context"
+    "fmt"
+    "log"
+
+    "github.com/tmc/langchaingo/llms/openai"
+)
+
+func main() {
+    ctx := context.Background()
+    
+    // ✅ Modern approach - uses gpt-4o-mini by default
+    llm, err := openai.New()
+    if err != nil {
+        log.Fatal(err)
+    }
+    
+    // ✅ If you need the old behavior
+    // llm, err := openai.New(openai.WithModel("gpt-3.5-turbo"))
+    
+    // ✅ If you want the best available model
+    // llm, err := openai.New(openai.WithModel("gpt-4o"))
+    
+    response, err := llm.Call(ctx, "What is the capital of France?")
+    if err != nil {
+        log.Fatal(err)
+    }
+    
+    fmt.Println(response)
+}
+```
+
+---
+
+## Milvus SDK Migration
+
+### What Changed
+
+The official Milvus Go SDK repository has been deprecated and archived. The new SDK is now part of the main Milvus repository.
+
+### Migration Steps
+
+#### 1. Update Dependencies
+
+In your `go.mod`:
+
+```diff
+- github.com/milvus-io/milvus-sdk-go/v2 v2.4.0
++ github.com/milvus-io/milvus/client/v2 v2.4.2
+```
+
+#### 2. Update Import Statements
+
+```diff
+- import "github.com/milvus-io/milvus-sdk-go/v2/client"
+- import "github.com/milvus-io/milvus-sdk-go/v2/entity"
++ import "github.com/milvus-io/milvus/client/v2/client"  
++ import "github.com/milvus-io/milvus/client/v2/entity"
+```
+
+#### 3. Run Migration Commands
+
+```bash
+# Update dependencies
+go get github.com/milvus-io/milvus/client/v2@latest
+go mod tidy
+
+# Test the changes
+go test ./vectorstores/milvus/...
+```
+
+### Files That Need Updates
+
+If you're using Milvus vector store directly:
+- Any files importing the old SDK
+- Integration tests
+- Example applications
+
+The API remains largely compatible, so most code should work without changes.
+
+---
+
+## Breaking Changes Checklist
+
+When migrating between versions, check for these common breaking changes:
+
+### v0.1.14+
+- [ ] Default OpenAI model changed to `gpt-4o-mini`
+- [ ] Review any hardcoded model assumptions
+- [ ] Test applications with new default behavior
+
+### v0.1.13+
+- [ ] Check Milvus vector store imports
+- [ ] Update deprecated Milvus SDK references
+
+### General Migration Tips
+
+1. **Pin your versions**: Use specific versions in production
+   ```go
+   // go.mod
+   github.com/tmc/langchaingo v0.1.13
+   ```
+
+2. **Test thoroughly**: Always test migrations in a staging environment
+
+3. **Monitor changes**: Subscribe to the [GitHub releases](https://github.com/tmc/langchaingo/releases)
+
+4. **Update incrementally**: Don't skip major versions if possible
+
+5. **Read release notes**: Check the changelog for each version
+
+---
+
+## Getting Help
+
+If you encounter issues during migration:
+
+1. Check the [GitHub Issues](https://github.com/tmc/langchaingo/issues)
+2. Search for existing solutions
+3. Create a new issue with:
+   - Your current version
+   - Target version
+   - Specific error messages
+   - Minimal reproduction code
+
+---
+
+## Version Compatibility Matrix
+
+| LangChainGo Version | OpenAI Default Model | Milvus SDK | Go Version |
+|-------------------|---------------------|------------|------------|
+| v0.1.14+ | gpt-4o-mini | New SDK | 1.21+ |
+| v0.1.13 | gpt-3.5-turbo | New SDK | 1.21+ |
+| v0.1.12 | gpt-3.5-turbo | Old SDK | 1.21+ |
+
+---
+
+*This migration guide is updated regularly. Last updated: 2025-08-29*