Public log

    Changelogwhat we shipped, when.

    Updated continuously. New skills, new models, new chains. Subscribe via RSS.

    Featurev1.4.0

    Per-mode landing pages, command palette and ⌘K everything

    • Five new dedicated landing pages — /nansen, /gmgn, /cloud, /bankr and /models — each with its own hero, capabilities, click-to-try queries, comparison table and FAQ.
    • ⌘K command palette — Linear/Raycast-grade, fuzzy ranked, grouped by Ask / Mode / Skill / Account / Navigate. Esc to close, ↑↓ to navigate, Enter to run.
    • Mobile bottom-sheet mode picker — replaces the horizontal scroll on small screens with a touch-friendly drawer.
    • Live market-pulse canvas behind the empty state, with prefers-reduced-motion respected.
    • First-visit cinematic intro that plays once and never gets in your way again.
    Skillv1.3.0

    GMGN-grade token alpha rebuilt on open APIs

    • Replaced the Cloudflare-blocked GMGN endpoints with a four-source aggregation: GeckoTerminal + DexScreener + GoPlus Security + RugCheck.xyz. Same on-chain data, no JS challenge.
    • Auto rug-check on every Solana result via RugCheck.xyz: bundling, mint authority, freeze authority, top-holder concentration.
    • GoPlus Security on every EVM token: honeypot test, buy/sell tax, blacklist, owner privileges.
    • Embedded GMGN trade buttons on every result so you click straight through to execute.
    Skillv1.2.0

    Nansen smart money, all 7 endpoints, no API key required

    • Conway now hits all seven Nansen smart-money endpoints from chat: DEX trades, Hyperliquid perps, Jupiter DCA, net flows, current and historical holdings, and the token screener.
    • x402 pay-per-call: roughly $0.01–$0.05 per query, billed from your Conway USDC balance. No Nansen subscription required.
    • Bring-your-own-Nansen-key option for users who already pay Nansen — Conway routes through your key for free.
    Modelv1.1.0

    13 frontier models, one flat price

    • Added GPT-5.2, GPT-5 Mini/Nano/Codex, GPT-4o + 4o Mini, Claude Opus 4.6 + 4.5, Claude Sonnet 4.5, Claude Haiku 4.5, Kimi K2.5, Kimi K2.5 (Nvidia), GLM-5 (Nvidia).
    • Same flat $0.015 per message regardless of model. Switch mid-conversation, no setting changes.
    • Side-by-side model comparison: 'Run GPT-5.2 and Claude Opus on this prompt' is now a first-class capability.
    Featurev1.0.0

    Conway public launch

    • Conway is live on Base. Pay-per-message in USDC, no subscription, $0.015 per message minimum.
    • Initial skill stack: Conway Cloud (Linux VMs), Conway Domains, Bankr trading (opt-in via API key).
    • Permissionless inference through Conway Tech infrastructure with sandboxed sessions.

    Want a feature?

    Ping us on Twitter or open an issue. The roadmap is set by what users actually use.

    Try Conway