Intent2Tx (Multi-Step): leaderboard
Metric: Final weighted score (from 0-1, times 100): per-instance 0.1 x JSON format validity + 0.3 x logic score (correct contract address, name and function) + 0.2 x parameter score + 0.4 x pass@1 (exact end-to-end match), averaged over instances, generating an Ethereum transaction (contract, address, function, parameters, ETH value) as JSON from a natural-language intent and an address book, on Intent2Tx instances derived from 300 days of real Ethereum mainnet traces, multi-step intents (an ordered transaction sequence), direct inference without demonstrations; higher is better. Source: arxiv.org. Saturation forecast: Around 2033. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.5 | 42 |
| 2 | Gemini 2.5 Pro | 38 |
| 3 | Kimi K2 0905 | 35 |
| 4 | Gemini 2.5 Flash | 33 |
| 5 | GPT-5.2 | 31 |
| 6 | Claude Haiku 4.5 | 31 |
| 7 | DeepSeek V3.2 | 31 |
| 8 | GLM-4.6 | 30 |
| 9 | Llama 3.3 70B Instruct | 29 |
| 10 | Qwen 3 235B A22B 2507 Instruct | 29 |
| 11 | codestral-2508 | 29 |
| 12 | Qwen3 Coder Next | 28 |
| 13 | GLM-4 32B (0414) | 27 |
| 14 | Llama 3.1 8B Instruct | 26 |
| 15 | GPT-5.2 Codex | 23 |
Interactive version: theaggregate.ai/benchmark?slug=intent2tx-multi-step · How It Works · Data refreshed daily, snapshot 2026-10-07.