multi-model-compare

GPT-4 vs Claude vs Gemini: Side-by-Side Evaluation

You need to choose between GPT-4, Claude, and Gemini for a task but have no data to compare them. This workflow sends the same prompt to all three providers in parallel via FORK_JOIN, collects their responses and token usage, and produces a side-by-side comparison.

Workflow

prompt
  │
  ▼
┌─── FORK_JOIN ────────────────────────────────┐
│                  │                            │
│ ┌──────────────┐ │ ┌──────────────┐          │ ┌──────────────┐
│ │mc_call_gpt4  │ │ │mc_call_claude│          │ │mc_call_gemini│
│ └──────────────┘ │ └──────────────┘          │ └──────────────┘
└────────┬─────────┴──────────┬────────────────┘
         ▼                    ▼
    ┌──────────┐
    │   JOIN   │
    └────┬─────┘
         ▼
  ┌──────────────┐
  │ mc_compare   │  Side-by-side comparison
  └──────────────┘
         │
         ▼
   comparison, winner

Workers

McCallGpt4Worker (mc_call_gpt4) -- Calls OpenAI Chat Completions with gpt-4o-mini, max_tokens: 512. Returns response text, model name, and token usage. In fallback mode, returns a deterministic response.

McCallClaudeWorker (mc_call_claude) -- Calls Anthropic Messages API with claude-sonnet-4-6. Returns response text, model name, and token usage. In fallback mode, returns a deterministic response.

McCallGeminiWorker (mc_call_gemini) -- Calls Gemini generateContent with gemini-2.0-flash. Uses contents[0].parts[0].text format. Returns response text, model name, and token usage. In fallback mode, returns a deterministic response.

McCompareWorker (mc_compare) -- Collects all three responses into a List.of(gpt4, claude, gemini) and produces a comparison summary.

Tests

14 tests cover all three provider calls and the comparison logic.

Further Reading

• RUNNING.md -- how to build and run this example