BREAKING: Baidu just open-sourced ERNIE 5.5 Flash Base.
This is a AI post classified by AGK Intelligence as Frontier models (a model release), kept by the AI Radar because it carries real work, not commentary.
BREAKING: Baidu just open-sourced ERNIE 5.5 Flash Base. 280B total / 14B active MoE. 33.5T pre-training tokens. Up to 1M context for fine-tuning. These are straight pre-train weights. No mid-training. No post-training. That is the product decision. You can measure what SFT and RL actually add from the same checkpoint, instead of reverse-engineering a chat model that already ate preference data. Same week as Kimi and Tencent Hy4. Open weights keep shipping. The scarce layer is moving to the recipe and the eval stack you put on top. If you fine-tune open models for a real product, what do y
Posted by Guillermo Flor (14k followers) 20 h ago · 1 likes · 383 views · view the original post on X. Kept by the AI Radar as Frontier models.
More AI work like this
- 🚨 STEALTH CHECKPOINT ALERT: Google is A/B testing Gemini 4 "Hydrogen" on LMArena… — @StatsWire
- Built a 100% procedural molten pāhoehoe lava simulation in Three.js using Gemini 4 Argon… — @thebuggeddev
- Coding tasks are saturated. So why is everyone still paying for the most expensive model… — @guilleflorvs
- What happens if Anthropic doesn't ship a new model for 6 months? — @guilleflorvs
- GPT-6 Cyber beats GPT-6 Astra on cybersecurity. — @adnaannasir_
- Can you see that dot right there in the top right? — @airesearch12
- Gemini 4 Hydrogen is currently testing on arena under the name gemini-3.8-flash and it's… — @thtbee_
- Gpt -6.2-Astra对比Gpt-6-Astra的前端之云上龙城：（从蛇变龙） — @leo114119
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 65.5k posts from 5k X accounts over the last 14 days, 10.4k tools, 19 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 16:50 UTC. Full method.