#416 Claude
A quick introduction to Claude and Claude Code; using ollama to swap models (including local models); and building a simple Chinese-English flashcard app as an example.
Notes
Anthropic was founded in 2021 by former OpenAI researchers. The company is structured as a Public Benefit Corporation with a mission to build “reliable, interpretable, and steerable AI systems” Key milestones:
- March 2023: Claude 1 launched (API + chat interface, limited access)
- July 2023: Claude 2 released—first public version with 100K context window
- March 2024: Claude 3 family (Haiku, Sonnet, Opus) introduced multimodal capabilities
- June 2024: Claude 3.5 Sonnet outperformed Claude 3 Opus on coding benchmarks
- October 2024: “Computer use” beta launched—Claude can control desktops
- May 2025: Claude 4 (Opus 4 + Sonnet 4) released with extended thinking mode
- May 2025: Claude Code moved from research preview to general availability
Product Components & Features
Clause has a web based UI for chat and research tasks.
Claude Code is installed locally:
- Terminal CLI
- IDE integration: VS Code, JetBrains
- Desktop app - standalone app for running Claude Code outside your IDE or terminal.
- Slack integration
Key technical features
- Context windows: 200K tokens standard (Haiku/Opus), with beta 1M token extension for Sonnet 4/4.5
- Multimodal: Text + image input capabilities
- Extended thinking: Reasoning mode that interleaves thinking with tool use
- Computer use: Ability to see screens, move cursors, click, and type across applications
- MCP connector: Model Context Protocol for tool integration
Claude Code
A command-line AI coding agent that lives in your terminal:
- Understands entire codebases
- Executes routine coding tasks via natural language
- Handles git workflows
- Native VS Code and JetBrains IDE integrations
- GitHub Actions support
- Checkpoints: Save progress and rollback capabilities (added in 2025)
- Acquired Bun in December 2025 to improve speed and stability
Most Suitable Use Cases
Software Development Content Creation & Communication Academic Research & Writing Enterprise Automation Specialized Applications
macOS Installation
$ curl -fsSL https://claude.ai/install.sh | bash
Setting up Claude Code...
✔ Claude Code successfully installed!
Version: 2.1.50
Location: ~/.local/bin/claude
Next: Run claude --help to get started
⚠ Setup notes:
• Native installation exists but ~/.local/bin is not in your PATH. Run:
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc && source ~/.bashrc
✅ Installation complete!
Nice! All installed, but not much use without a subscription.
Using Claude with a model served by Ollama
https://docs.ollama.com/integrations/claude-code
Recommended models:
- qwen3-coder
- glm-4.7
- gpt-oss:20b
- gpt-oss:120b
Claude Code requires a large context window. I’ve set the ollama context window to 64k tokens.
Using qwen3-coder, for example, by pulling the model locally and specifying it for use with claude:
ollama pull qwen3-coder
ANTHROPIC_AUTH_TOKEN=ollama ANTHROPIC_BASE_URL=http://localhost:11434 ANTHROPIC_API_KEY="" claude --model qwen3-coder
Building a Flashcard App
Let’s build a little Chinese-English flashcard app with claude.
I added a basic CLAUDE.md and then build the app in three prompts:
create data.json with at least 100 examples create index.html app with css and js assets create sinatra stub app.rb with Gemfile and .ruby-version and .ruby-gemset
I had to futz around a bit to get it to correct the animation sequencing (initially it didn’t allow for css transforms to complete before accidentally showing the answers), but now I have a nice little app. Source files:
There’s also a sinatra wrapper so I can run it locally:
cd flashcards
ruby app.rb
In action, the main screen:
Showing a question:
And revealing the answer:
