FILTERED RESULTS
FILTERS
Ads Top
DARK MODE
CHART
MCap $2.9T +0.5%24h Vol $93.5B -13%Fear & Greed 74/100Alts Index 53/100
BTC.D 58.4% 0%Stable.D 9.2% 0%ETH.D 11.4% 0%Others.D 21.0% 0%
CAP$0.0819+29.26%•BP$1.487+23.22%•BTW$1.422+22.6%•MON$0.0324+17.18%•STX$0.3761+15.19%•JASMY$0.00580038+14.35%•TRAC$0.4476+11.3%•VELO$0.0056984+11.18%•PURR$0.1621+10.81%•HUMA$0.0330+8.25%•
STONK$0.2313-20.63%•AI$0.1487-17.91%•2Z$0.0574-8.62%•SOON$0.4265-8.28%•GRASS$0.6671-8.28%•MET$0.2969-8.15%•KMNO$0.0409-7.95%•TIBBIR$0.2716-7.85%•QNT$265.98-7.46%•RAY$1.898-6.68%•
Top movers 24h
    Filters
      Coins
      Sentiment
      Impact
      Search
      FILTERED RESULTS

        

      Upgrade your plan
      Dashboard

      Google Prices Gemini 4 Argon at $2/$10, the Same Rate as…

      Google released Gemini 4 Argon on 30 September at an introductory price of $2 per million input tokens and $10 per million output tokens, matching the exact rate Anthropic charges for Claude Sonnet 5.5 and OpenAI charges for GPT-6.1 Sol.The model launched exclusively to trusted cybersecurity testers through Google's Fairwind Program, with no public access date announced.The pricing convergence means three competing frontier labs now sell their latest-generation models at identical per-token rates, a development that compresses the margin available to any single provider and signals that API pricing has become a commodity lever rather than a differentiator. Google's standard post-promotional pricing for Argon rises to $4 per million input tokens and $20 per million output tokens, according to 9to5Google, which matches Anthropic's Claude Opus 5.5 tier.

      Google's Benchmark Claims and the Accuracy Question

      Google said Argon is "fundamentally changing the way we work and build at Google," according to TechCrunch. The company claims the model scored 77.9% on DeepSWE v1.1 and ranked first on Zapier's AutomationBench at 51.3%, according to figures Google published alongside the release. On the Vals Index, Google reported 68.9% compared to Claude Opus 5.5's 67.0%, per Cyber Kendra. Independent testing by Artificial Analysis found that Argon gave wrong answers on 15% of AA-Omniscience questions and had raw accuracy of 50%, five percentage points below Gemini 3.1 Pro Preview, according to Cyber Kendra.Google cancelled the previously scheduled Gemini 3.5 Pro before pivoting to the Argon architecture, suggesting the development path was not straightforward. The model also trails OpenAI's GPT-6 Astra on FrontierSWE v2, scoring 55.0% against Astra's 65.5%, based on Google's own published benchmark results.Tulsee Doshi, Product Lead at Google, said the team is "putting a strong defensive tool in the hands of security experts first" to give Google time to strengthen safeguards against prompt injections and misalignment, according to Android Headlines.The staged rollout means Argon is initially limited to trusted cybersecurity testers, while Google's release information says access will expand to Google AI Ultra subscribers and paid API customers.

      Alphabet Stock Moves on the Release

      Alphabet shares closed at $344.08 on 30 September, up 0.93% during the regular session. The stock rose about 1% to 2% in after-hours trading, according to reports.The after-hours move suggests investors viewed the Argon release as incrementally positive, though the gains remained modest relative to the scope of the product announcement. Argon's technical specifications include a one-million-token output limit, up from 64,000 in the prior generation. A Long Decode Continuation feature that allows pausing and resuming inference across calls. Koray Kavukcuoglu, Senior Vice President of Google DeepMind, said the expanded output window "adds a new level of depth in reasoning," according to 9to5Google. Internally, Google reported that Argon freed more than 300 terabytes of data center memory and is being used to migrate over 800,000 lines of Fuchsia Zircon kernel code from C and C++ to Rust.

      What to Watch Out For

      Google has not disclosed when Argon will become available to general API customers or consumer products such as Google Search and Workspace. The company says the model will roll out next to Google AI Ultra subscribers and paid API customers, but has not provided a broader public availability timeline.The gap between the cybersecurity-only launch and public availability will determine whether the introductory $2/$10 pricing attracts meaningful developer adoption before the promotional window closes and standard rates double to $4/$20 per million tokens.Developers building on Google's API must also weigh the 15% share of wrong answers on Artificial Analysis's AA-Omniscience questions against the model's strong automation and coding benchmarks when deciding whether to migrate from Gemini 3.1 Pro or switch from a competing provider.

      Source: FinanceFeeds
      .

      Terra Founder Do Kwon Sentenced to 15 Years in Prison for Fraud