Releases·Model Releasefeatured
Google announces Gemini 4 Argon AI model amid employee skepticism, but you can't use it yet
Google DeepMind launched Gemini 4 Argon on September 30, 2026, a frontier model built for deep reasoning across coding, enterprise knowledge work, and cybersecurity defense. Access is limited to government users and vetted cyber defenders in the Fairwind Program, with broader rollout promised after safety refinement. Google claims first place on 13 of 19 benchmarks against GPT-6 Astra and Claude Opus 5.5, though independent evals show a narrower picture and some observers question the published numbers.
3688 engagement15 sourcesWed, Sep 30, 2026, 08:11 PM
Developments
Deep Think rollout announced; users expect benchmark retests, noting Argon's coding performance lags text, math, and finance.
Sun, Oct 4, 2026, 05:41 PM
Deep Think is being rolled out again for Gemini 4 Argon; users ask whether it will lift coding benchmark results tested at High.
Sun, Oct 4, 2026, 05:41 PM
AINews adds pricing ($2/$10 discounted), Long Decode Continuation detail, third-party evals, and benchmark skepticism
Fri, Oct 2, 2026, 06:24 PM
AI Overview
- Google claims Gemini 4 Argon leads in coding, knowledge work, and cybersecurity, scoring 77.9% on DeepSWE v1.1 above GPT-6 Astra, Fable 5.1, and Opus 5.5.
- Internal Argon agents saved Google 300 TiB of data-center memory via fleet-wide telemetry and migrated 800,000+ lines of the Fuchsia Zircon kernel to Rust.
- Google released Gemini 4 Argon on October 1, 2026; GDM last shipped a larger-than-Flash model in February (3.1 Pro), and Argon enters limited cybersecurity preview with broader access promised 'as soon as possible'.
- Argon agents migrated thousands of lines of the re2 and libgav1 C/C++ libraries to Rust as part of Google's codebase migration effort.
- Google cites Argon's industry-leading score on the economic analysis Vals Index as evidence of strength on long-horizon tasks.
- Bloomberg reports Google employees voiced skepticism internally about the Gemini 4 Argon model's announcement.
- Google limited initial Gemini 4 Argon access to 'trusted cyber defenders' while engaging the U.S. government's voluntary pre-release model access process; Koray Kavukcuoglu called it a 'phased approach' to releasing frontier capabilities.
- Google DeepMind's Argon hits 1M output tokens via the new Long Decode Continuation API feature, up from 64K; Vals lists 262K max output.
Entities
Related
Social·Communityfeatured
Opus 5.5 Still Overthinks at Higher Effort
After Anthropic's Opus 5.5 and OpenAI's GPT-6 Sol launched the same day, users compiled Artificial Analysis and announcement-chart comparisons. Opus 5.5 wins raw benchmark scores, but GPT-6 Sol delivers far better score-per-dollar and speed, with low-reasoning Sol matching last-generation medium-level performance at roughly a quarter of the cost. Follow-up analyses extend the picture to Sonnet 5.5 at matched token prices and flag Grok 4.7's high per-task cost.
After Anthropic's Opus 5.5 and OpenAI's GPT-6 Sol launched the same day, users compiled Artificial Analysis and announcement-chart comparisons. Opus 5.5 wins raw benchmark scores, but GPT-6 Sol delivers far better score-per-dollar and speed, with low-reasoning Sol matching last-generation medium-level performance at roughly a quarter of the cost. Follow-up analyses extend the picture to Sonnet 5.5 at matched token prices and flag Grok 4.7's high per-task cost.
Updated Thu, Oct 1, 2026, 05:05 PM — Sonnet 5.5 comparison at matched token prices: Sol more efficient per task, Sonnet 5.5 higher ceiling (56 vs Sol max 48)
1141.5k engagement11 sourcesredditTue, Sep 22, 2026, 07:21 PM
Releases·Model Releasefeatured
OpenAI launches GPT-6 Sol and Luna at half the cost of 5.6 series
OpenAI released GPT-6 Sol and GPT-6 Luna on September 22, 2026, bringing GPT-6 Astra's advances in coding, factuality, computer use, and agentic work into faster, cheaper models. Sol costs $2/M input and $10/M output, Luna $0.10/M and $0.50/M — about 50% below GPT-5.6 promotional pricing. The launch came the same day as Anthropic's Claude Opus 5.5 and immediately reshaped the frontier pricing landscape.
OpenAI released GPT-6 Sol and GPT-6 Luna on September 22, 2026, bringing GPT-6 Astra's advances in coding, factuality, computer use, and agentic work into faster, cheaper models. Sol costs $2/M input and $10/M output, Luna $0.10/M and $0.50/M — about 50% below GPT-5.6 promotional pricing. The launch came the same day as Anthropic's Claude Opus 5.5 and immediately reshaped the frontier pricing landscape.
Updated Thu, Sep 24, 2026, 03:59 AM — September 24 independent-testing coverage of the GPT-6 family reports benchmark controversy alongside confirmed launch f
96060.6k engagement16 sourcesrss, reddit, hackernews +2Tue, Sep 22, 2026, 12:00 AM
Social·Community
GPT-6 Astra vs Claude Opus 5.5: comparison splits on coding, price, security, context, and reasoning
A practitioner compared GPT-6 Astra (released Sept 3, 2026) and Claude Opus 5.5 (released Sept 22, 2026) head-to-head on coding, price, security, context, and reasoning. Opus 5.5 wins coding and cost; Astra leads security, reasoning, and context. The takeaway: the models suit different jobs rather than one dominating.
A practitioner compared GPT-6 Astra (released Sept 3, 2026) and Claude Opus 5.5 (released Sept 22, 2026) head-to-head on coding, price, security, context, and reasoning. Opus 5.5 wins coding and cost; Astra leads security, reasoning, and context. The takeaway: the models suit different jobs rather than one dominating.
12 engagement1 sourceredditWed, Sep 23, 2026, 02:37 PM
Developments·Benchmark
Gemini 4 Argon beats GPT-6 Astra and Claude Fable 5.1 on Google's published benchmark table
Google published a benchmark table on September 30 showing Gemini 4 Argon beating GPT-6 Astra and Claude Fable 5.1 on most benchmarks. A Towards AI article argues the top performer is the one buyers can't actually purchase. For practitioners, leaderboard results may not reflect accessible model availability.
Google published a benchmark table on September 30 showing Gemini 4 Argon beating GPT-6 Astra and Claude Fable 5.1 on most benchmarks. A Towards AI article argues the top performer is the one buyers can't actually purchase. For practitioners, leaderboard results may not reflect accessible model availability.
0 engagement1 sourcerssThu, Oct 1, 2026, 03:43 PM