ERNIE 5.5 Flash Preview: Baidu's 280B MoE with 1M context
Baidu's first post-trained ERNIE 5.5 Flash checkpoint: a 280B-total / 14B-active sparse MoE with a 1M-token context, live on Qianfan and free on OpenRouter, with Apache 2.0 weights promised by 25 Nov.
Why it matters
A 280B model that only activates 14B parameters per token, serving 1M-token contexts at $0.28/$1.13 per million tokens — Baidu's cheapest long-context frontier-class model, with open weights due within days.
What you could build with it
A small team building a long-document review tool could route 1M-token contract or research-paper workloads through the free OpenRouter tier to test whether the cheap long context holds up on real documents before committing to a paid provider.
Does it hold up?
Too early to judge: the API is live and free on OpenRouter, but the weights are not out yet, there are no independent benchmark scores, and Baidu itself calls it an intermediate checkpoint that may have capability gaps.
Built with ERNIE 5.5 Flash Preview
- How to Use ERNIE 5.5 Flash API with Qianfan and OpenRouterapidog · Step-by-step guide to calling the model on Qianfan and OpenRouter, with the model IDs and limits checked against the official pages.
- ERNIE 5.5 Flash Preview: API Today, Open Weights Before Nov 25modelfit · Checks what Baidu actually confirmed versus promised, sizes the FP8 weights at about 288 GB, and notes no runtime supports the architecture yet.
- What Is ERNIE 5.5 Flash? Specs, Pricing & Benchmarkskie.ai · Collects the official specs, the $0.28/$1.13 pricing and the 200+ tok/s throughput claim, with the Nov 25 weights date.
Learn more
First spotted on article: source.
More AI models
Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Qwen4-Max: Alibaba's new flagship model goes liveAlibaba's 2.4T-parameter flagship is now generally available via API, with a 1M-token context, native vision, and…models · JEV 0.67smolmaxxing: run 16B LLMs on a 1 GB TV box or a 2013 iPadAn MIT-licensed llama.cpp patch set that runs language models up to a 16B MoE on devices with about 1 GB of usable RAM…models · JEV 0.65Blue TTS v3: a 388MB open text-to-speech model with voice cloningAn open-source 388MB text-to-speech model (MIT) with voice cloning from about 10 seconds of speech, 7 languages, 15…models · JEV 0.65Haru-4BOpen full-duplex spoken-dialogue foundation model that listens and speaks simultaneously (7.5 Hz frames, 133 ms per…models · JEV 0.63
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.