OpenRouter reported DeepSeek V4.1 Flash processed ~1T tokens in 24 hours and is on pace for ~2.8T in 48 hours—potentially its biggest paid-model launch window. About 90% of those tokens were cache reads, priced by the market at ~$0.006/M, roughly 5x cheaper than comparable models like GLM-5.3 Flash.
OpenRouter’s Jeff Wang said Fable 5.1 is smarter with much better cache-hit economics, so they can offer it at under half the price of Fable 5 while improving fusion routing. That shows gateways can pass Anthropic’s cache-read cut through as an immediate end-user price win.
OpenRouter also announces Muse Spark 1.3 Contributor is live as the cost-efficient tier for experimentation, learning, and early agentic/coding workflows. Listed around $0.10 in / $0.20 out per 1M tokens (vs ~$1.25/$4.25 on the standard tier); prompts and outputs may be used to improve Meta products.
Abacus AI CEO @bindureddy announced open-weights Smaug Flash, a personal-agent fine-tune on RouteLLM API at about $0.10/M input and $0.40/M output—claiming DeepSeek Flash-level performance at roughly 4x lower price, with weights on Hugging Face.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.