OpenRouter launches Perceptron Mk1 with 33k context at $0.15/M input, $1.50/M output
TL;DR
OpenRouter launched Perceptron Mk1, a vision-language model for video and embodied reasoning. It processes images and videos with 33k context at $0.15/M input and $1.50/M output.
What changed
Perceptron Mk1 is now available on OpenRouter. This vision-language model handles video and embodied reasoning with 33k context length. It processes image and video inputs alongside natural language queries for detailed visual understanding, priced at $0.15 per million input tokens and $1.50 per million output tokens.
Why it matters
Developers integrating video analysis gain a cost-effective option at $0.15/M input versus GPT-4V's higher rates on the same platform. Vibe Builders can generate detailed insights from video clips for creative projects. The 33k context supports longer sequences in embodied reasoning tasks.
What to watch for
Compare output quality against LLaVA models on OpenRouter for video queries. Test Perceptron Mk1 directly via the OpenRouter playground with a sample video input. Monitor usage limits and rate card updates on the model page.
Who this matters for
- Vibe Builders: Use Perceptron Mk1 to extract detailed visual insights from video clips for your creative projects.
- Developers: Integrate this cost-effective vision model for video analysis tasks at $0.15 per million input tokens.
Harsh’s take
Perceptron Mk1 enters a crowded vision-language market with a clear pricing advantage. By undercutting premium models on OpenRouter, it forces a shift in how teams approach video-heavy workflows. The 33k context window is sufficient for most short-form video reasoning tasks, making it a practical choice for developers who need to balance performance with operational costs.
Success depends on the model's ability to maintain accuracy during complex visual reasoning. Builders should prioritize benchmarking this model against existing LLaVA variants to determine if the cost savings justify a migration. If the reasoning quality holds up, this becomes a primary tool for high-volume visual data processing.
Focus on testing specific video input types to verify if the model meets your production requirements.
by Harsh Desai
About OpenRouter
View the full OpenRouter page →All OpenRouter updatesGo deeper
More AI news
- LaunchAsian AI startups launch Mythos-like models as Anthropic export ban continues
Asian AI startups launched models with Mythos-like capabilities. The releases follow Anthropic's ongoing export restrictions.
- Daily RoundupGemini jetlag aid, OpenAI Jalapeño chip, and Vercel agent tools (daily focus hooks)
Google, Vercel, and OpenAI shipped practical AI updates while new models and benchmarks highlighted shifting hardware and capability limits.
- Model ReleaseOpenAI limits GPT-5.6 rollout after government request, says restrictions shouldn’t be the norm
OpenAI limited GPT-5.6 rollout after a government request. The company stated that such restrictions should not become the long-term default.