Releases·Model Releasefeatured
Google launches Gemini 3.8 Flash and 3.8 Flash Cyber
Google released Gemini 3.8 Flash, publishing a model card on DeepMind's site alongside a companion Flash Cyber variant. The launch drew heavy discussion on Hacker News and coverage from The Register, and the model reached GitHub Copilot the next day.
1851 engagement6 sourcesWed, Sep 2, 2026, 03:12 PM
Developments
Gemini 3.8 Flash is now available in GitHub Copilot (2026-09-03).
Thu, Sep 3, 2026, 07:22 PM
AI Overview
- Google announces Gemini 3.8 Flash on a DeepMind page (September 2, 2026), the news topping hackernews
- Google's announcement page gives no model details in the linked text
- A Gemini 3.8 Flash model card PDF (hosted on Google storage) circulates via hackernews the same day
- Google's announcement names a second model alongside Flash, Gemini 3.8 Flash Cyber
- The Register's coverage frames the launch as Google signaling it remains competitive in the AI race
- GitHub Copilot adds Gemini 3.8 Flash (github.blog, September 3, 2026)
Related
Social·Community
Google lost its AI lead in 7 months: no Gemini 3.5 Pro, talent exodus, and TPUs handed to rivals
Reddit threads debate why Google lost AI momentum despite unmatched resources, citing no new Gemini Pro model since Gemini 3.1 Pro topped the Artificial Analysis index in February 2026, executive departures, and TPU capacity routed to rivals. A follow-up post reports the Gemini 4 Argon announcement excluded paid Pro-tier users, prompting subscription cancellations.
Reddit threads debate why Google lost AI momentum despite unmatched resources, citing no new Gemini Pro model since Gemini 3.1 Pro topped the Artificial Analysis index in February 2026, executive departures, and TPU capacity routed to rivals. A follow-up post reports the Gemini 4 Argon announcement excluded paid Pro-tier users, prompting subscription cancellations.
Updated Thu, Oct 1, 2026, 03:41 PM — Gemini 4 Argon announced with no access for Pro-tier users, limiting it to Fairwind partners, paid API, and a new Ultra
001391 engagement3 sourcesredditSun, Sep 27, 2026, 04:57 PM
Social·Community
Redditor stress tests Gemini 3.8 Flash in GitHub Copilot; it fails while GPT 6.1 Sol and Sonnet 5.5 hit 94% accuracy
A developer stress-tested five models in GitHub Copilot on a legal/compliance documentation task using identical inputs (23 files, 65k chars). GPT 6.1 Sol and Sonnet 5.5 hit 94% accuracy, while Gemini 3.8 Flash High failed to produce the output document and burned 400 AI credits before being stopped. The poster concludes Gemini models remain fine for chat but are risky for dev tasks.
A developer stress-tested five models in GitHub Copilot on a legal/compliance documentation task using identical inputs (23 files, 65k chars). GPT 6.1 Sol and Sonnet 5.5 hit 94% accuracy, while Gemini 3.8 Flash High failed to produce the output document and burned 400 AI credits before being stopped. The poster concludes Gemini models remain fine for chat but are risky for dev tasks.
16 engagement1 sourceredditWed, Sep 30, 2026, 02:16 PM
Releases·Model Release
Z.ai releases GLM-5.3-Flash, a new model in the GLM series
Z.ai has released GLM-5.3-Flash, a new model in the GLM series, with the announcement shared on Hacker News and Hugging Face. The model's page is live on Z.ai's website, and early analysis from Artificial Analysis covers its intelligence, performance, and pricing. Specific technical details are not yet available from the posts.
Z.ai has released GLM-5.3-Flash, a new model in the GLM series, with the announcement shared on Hacker News and Hugging Face. The model's page is live on Z.ai's website, and early analysis from Artificial Analysis covers its intelligence, performance, and pricing. Specific technical details are not yet available from the posts.
3202.7k engagement5 sourceshackernews, redditWed, Aug 26, 2026, 02:08 PM
Social·Community
Deprecating Gemini 3.7 makes zero sense.
A developer reports daily use of Gemini 3.7 Flash Medium for code edits, arguing Google's deprecation of the model is misguided. They describe 3.8 Flash as slower and more expensive on small tasks, spending extra reasoning tokens on trivial fixes. The post reflects practitioner concern that model retirements can remove fast, cheap options for lightweight coding workloads.
A developer reports daily use of Gemini 3.7 Flash Medium for code edits, arguing Google's deprecation of the model is misguided. They describe 3.8 Flash as slower and more expensive on small tasks, spending extra reasoning tokens on trivial fixes. The post reflects practitioner concern that model retirements can remove fast, cheap options for lightweight coding workloads.
4 engagement1 sourceredditSun, Oct 4, 2026, 12:27 PM
Releases·Model Release
Google's Gemini 4 Argon post on blog.google hits 983 points on Hacker News
Google launches Gemini 4 Argon
Google launches Gemini 4 Argon
2.7k engagement1 sourcehackernewsWed, Sep 30, 2026, 08:04 PM