SKIP TO WAITLISTX / TWITTER

Duel Agents is an IDE-native framework that runs every model you have access to against each other on the same prompt. The cheapest answer that still passes our quality checks wins.

Flagship-level results, without flagship-level bills.

/ ROUTE/ SCORE/ SPLIT
01 · WHAT

PROMPT → MODELS → WINNER → SUB → MINI

PROMPTMODEL AMODEL BMODEL CMODEL DMODEL n+1SUBSUBSUBSUBMINI-AGENTS
02 · HOW

−73% SPEND VS FLAGSHIP-ONLY, SAME TASK SET
FLAGSHIP-ONLYDUEL AGENTS
$0.20$0.15$0.10$0.05$0.00
03 · PROOF

  • / NATIVE

    Plugs into Cursor, Claude Code, Codex, OpenClaw and other agent IDEs as a drop-in router. No new shell, no new contract.

  • / MERIT

    Models are picked by public benchmarks plus our private eval suite, scored in real-time on the actual task you sent.

  • / SPLIT

    Recursive decomposition: jobs fan out into sub-agents, sub-agents into mini-agents on the cheapest tier that still wins.

04 · WHY
03 · ECOSYSTEM
Ethereum
Solana
USDC
USDT
/ PAYMENTS

Pay with any crypto. Ethereum, Solana, USDC, USDT, and more.

NOW SHOWING ·
ChatGPT
Claude
Gemini
DeepSeek
Mistral
Llama
/ MODELS

Duel runs ChatGPT, Claude, Gemini, DeepSeek, Llama, Mistral, Perplexity. Whichever wins your task.

NOW SHOWING ·
Venice
Nous Research
OpenClaw
Cursor
Codex
Claude Code
/ PLATFORMS

Use Duel with Claude Code, Venice, Nous Research, OpenClaw, Cursor, Codex.

NOW SHOWING ·
04 · REQUEST ACCESS

Get in early.

We're onboarding a small first wave. Drop your email; we send one message when access opens.

· NO MARKETING · ONE EMAIL WHEN ACCESS OPENS ·

Frame 1 of 4: Landing