An AI-native media engine from MIT computer science graduates and YC S26, designed to replace fragmented creative tools with a single conversational workspace. Instead of training its own models, Palette routes requests across the best third-party engines—ByteDance Seedance 2.5, Google Veo 3, MiniMax H3, Alibaba Wan 3, and others—picking the right one for the job (e.g., physics for Seedance, face consistency for MiniMax, realism for Veo) . It unifies generation, editing, storyboarding, character consistency, and brand kits into one canvas, and exposes the engine via REST API and MCP Server for use in Claude Code or custom apps . Pricing is unusually team-friendly: Free (€0, 200 credits), and Team at €149/month for unlimited users and unlimited credits (introductory deal, bring your own Claude/Codex subscription) . Caveats: very early-stage (launched July 2026, team of 2), limited independent user reviews, and the flat-rate intro pricing may change.
Pros & Cons:
Pros: The model routing approach is genuinely useful—instead of subscribing to Runway, Kling, and Veo separately, you describe the shot and Palette picks the engine that fits (or lets you choose manually), which reduces cost and decision fatigue for teams that need variety ; the flat org-wide pricing (€149/mo unlimited users) is a strong differentiator for agencies and marketing teams, avoiding the per-seat traps of most SaaS tools ; it covers the full workflow—character consistency, storyboarding, brand kits, and auto-repurposing—rather than just text-to-video; the API and MCP support make it usable inside developer and AI-assistant workflows .
Cons: It is extremely early-stage—founded in 2026, team of 2 (MIT grads with strong backgrounds), and no large-scale independent user reviews surfaced, so reliability and support capacity are unproven ; the “unlimited credits” Team pricing is an intro deal, and it is unclear what happens when the deal ends (potential for price hikes or reversion to metered billing) ; it is a wrapper around third-party models, so its output quality is ultimately bounded by those models, and its value depends on whether the routing and workflow layer is actually better than manually using those models; the product surface is ambitious (generation, editing, storyboards, enterprise workflows, API), and for a 2-person team, feature depth and stability may lag mature competitors.
Who is it for?:Marketing teams, agencies, and enterprises that produce high volumes of AI video and want to unify model access, brand consistency, and workflow in one place; creators who already pay for Claude/Codex and want to generate media without leaving their AI assistant (via MCP); teams that value flat, predictable pricing over per-seat or per-credit costs; developers who want API access to multiple video models through one endpoint. Not suited for individual creators seeking a simple, polished one-click tool (Runway/Pika are more direct), teams that require extensive independent user reviews before adopting a new platform, or anyone who needs a self-hosted or open-source solution (Palette is proprietary and cloud-based).