Early hands-on: using Astra to update research papers, with takes on Opus 5.5 and Gemini 4
JasonBotterill · x · 2026-10-26
JasonBotterill shares a hands-on test: last week he went over random papers to see whether Astra could produce a meaningful update for a researcher — and it could.
His impressions of the models involved:
- Opus 5.5 has really good "taste" and writes in a much more palatable way.
- Gemini 4 is amazing at search.
- Astra is "autistic (good)" and has a critical lens.
Third-party impressions of reportedly unreleased models; unverified.
Related event: Test of Unreleased Models on Random Papers: Opus 5.5 Shows Great Taste(2 posts)→
More from Models
- DeepSeek 1017 checkpoint lands on FrontierCS 2 with a small score jump — scaling01 · 2026-10-26
- Opus 5.5 hailed as best model ever yet wildly inconsistent, dev complains — ivan_bezdomny · 2026-10-26
- Developer: Claude models make 'crazy mistakes' after just two days away — ivan_bezdomny · 2026-10-26
- Gemma 4 creative writing woes: every 26B finetune makes characters sound identical — Whydoiexist2983 · 2026-10-26
- Tencent open-sources WeVisOne-VL: one VLM for detection, OCR, segmentation and matting — lmoroney · 2026-10-26
- New model weight sizes: Q4_K_M GGUF fits a 24GB card with ~20.8GB — AIFlow_ML · 2026-10-26