This is Step 5 Preview in comparison with other frontier models.
This is a AI post classified by Jev as Frontier models (a model release), kept by the AI Radar because it carries real work, not commentary.
This is Step 5 Preview in comparison with other frontier models. A 600B model with only 27B active params, and the early numbers look insane: > 67.7 on DeepSWE > Beats Gemini 3.8 Flash > Outperforms most Chinese models > 1M context window with image and video input > $1 input / $2.70 output It's already live in the API and chat, so you can try it now. would love to get acces and test it against a closed source model like Astra :)
Posted by Veee (9.1k followers) 1 h ago · 14 likes · 604 views · view the original post on X. Kept by the AI Radar as Frontier models.
More AI work like this
- Fable 5.2 : My Impressions — @pankajkumar_dev
- So, looks to be another preview-then post train further. At 600b as an moe this is good… — @DJLougen
- Me watching Fable one-shot the whole repo into Rust. — @dedene
- there is no way this is real — @notjazii
- Anthropic cooked with Opus 5.5 🔥🔥 — @HarshithLucky3
- first we got Jev from TypeSafe AI, a model that's insanely cheap! — @vikktorrrre
- Fable 5.2, Gemini 4 Pro, and GPT-6 Astra (Pro) competing for the best SVG. — @TokenGremlin
- 🚨 Gemini 4 Pro vs Fable 5.2 vs Astra pro (Xbox controller SVG) — @LuminaBench
Every post is read and classified by Jev (TypeSafe): what it is, which market it belongs to, and whether the link is a real tool. 34.4k posts from 4.9k X accounts over the last 14 days, 1.5k tools, 19 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-20 13:27 UTC. Full method.