AI tools: Claude & Gemini
This course compares how Claude and Gemini handle reasoning on a large, tangled coding task. It explains what a thinking mode returns in each API and how both providers bill thinking tokens. It compares Claude's effort parameter with Gemini's thinking level, and checks both context windows against current model pages. It covers the cost and latency of deeper reasoning. It then applies these settings to Enrollment::UserGamificationQuery, a 48-line query with seven collaborators, and closes with choosing reasoning depth before reviewing the answer.
What you'll learn
- ✓What thinking blocks and thought steps look like in Claude and Gemini API responses
- ✓How Claude's effort and Gemini's thinking_level steer reasoning depth
- ✓How context windows and output limits compare across both providers' newest models
- ✓What deeper reasoning costs in tokens and latency
- ✓How to match reasoning depth to the shape of a task
- ✓Why a well-reasoned answer still needs review against the code
From directing one tool to choosing between two
By the end of the intern-level AI courses, "reach for an agentic tool" meant something specific. You opened Cursor or Antigravity, pointed it at a request, and watched it read a file, write code, run a command and read the result. "AI tools: introduction (Cursor & Antigravity)" checked that loop…
Read the full lesson free →Curriculum
7 lessonsFrom directing one tool to choosing between two
Introduces choosing between Claude and Gemini for reasoning, using a query with seven collaborators as the example task.
What a thinking mode does inside a model
Shows how Claude returns thinking blocks and Gemini returns thought steps, and how both bill thinking tokens.
Two dials for the same idea: effort and thinking level
Compares Claude's effort parameter and Gemini's thinking_level, their defaults per model, and the older numeric budget modes.
Context windows: both lineups now in the same range
Checks current context windows and output limits for Claude and Gemini, and separates window size from reasoning depth.
Reasoning depth is a cost and latency dial
Explains how each provider frames reasoning depth as a tradeoff against cost, latency and capability, with guidance per task.
Sizing up a query with seven collaborators
Reads the collaborators and edge cases of UserGamificationQuery to show why this task calls for a higher reasoning setting.
Putting it together: picking reasoning depth before you review
Summarizes the comparison and frames the decision as matching reasoning depth to a task, followed by review.