Observed Signal · Jul 21, 2026 · Product Launch · Source: CNBC Technology · Impact: 4/5 · Sentiment: Positive
Google expands Gemini with cheaper models, Mythos rival
Google DeepMind released three new Gemini models—Gemini 3.6 Flash, 3.5 Flash‑Lite and 3.5 Flash Cyber—aimed at users building and running AI agents with improvements in efficiency, latency and reliability. Gemini 3.6 Flash is positioned as the workhorse: it reportedly uses about 17% fewer tokens than its predecessor, boosts coding, multimodal and knowledge‑work performance, and Google says it is cheaper per task than some competing offerings. Flash‑Lite is the fastest, most cost‑efficient option for high‑volume or cost‑sensitive workloads. Flash Cyber is fine‑tuned to find and patch security vulnerabilities and will initially be available only to governments and trusted partners via a limited pilot. Google is also testing Gemini 3.5 Pro with partners (launch delayed for performance fixes), developing a specialized chip to run Gemini more efficiently, and has begun a major pretraining run for Gemini 4.
Major-platform technical model launches from Google/Alphabet affect model cost, competitive positioning vs. Anthropic/OpenAI/Chinese rivals, and have implications for infrastructure and AI serving economics.
Track Google Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Google released three new Gemini models: 3.6 Flash, 3.5 Flash‑Lite and 3.5 Flash Cyber.
- Gemini 3.6 Flash cuts token usage by about 17%, improves coding, multimodal and knowledge‑work performance, and is claimed to be cheaper per task than some rivals (e.g., OpenAI, Moonshot AI, Alibaba).
- Gemini 3.5 Flash‑Lite is positioned as the fastest, most cost‑efficient option for high‑volume or cost‑sensitive workloads.
- Gemini 3.5 Flash Cyber is tailored to detect and patch software vulnerabilities and will be available initially only to governments and trusted partners via a limited pilot.
- Google is testing Gemini 3.5 Pro with partners (launch delayed due to performance issues), is developing a specialized chip to run Gemini more efficiently, and has started a major pretraining run for Gemini 4.
Connected Companies & Entities
7 Entities mapped“Google is also launching Gemini 3.6 Flash, which improves coding, multimodal and knowledge-work performance while using up to 17% fewer toke...”
“Alphabet is releasing three new Gemini models on Tuesday, including its clearest answer yet to Anthropic’s lead in cybersecurity, as the com...”
“A Google Cloud spokesperson told CNBC in a statement that its teams are 'constantly researching and experimenting with new innovations to de...”
“The new model could help Google narrow its cybersecurity gap with Anthropic, which has built an early lead in automated code defense with it...”
“Artificial Analysis data shows Gemini Flash already undercuts comparable models from Anthropic, OpenAI and Chinese rivals on cost....”
“Moonshot AI’s Kimi K3 drew enough demand that the company limited new subscriptions and API access because of capacity constraints....”
“Alibaba is teasing Qwen 3.8 Max, which it said trails only Anthropic’s Fable 5 in overall performance....”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Google Launches Gemini 3.8 Flash and Cyber AI Models
Google launched Gemini 3.8 Flash and Gemini 3.8 Flash Cyber on September 3, 2026, both derived from the same foundation model. Flash is optimized for coding, reasoning, and AI agents, while Flash Cyber specializes in defensive cybersecurity and vulnerability patching. The models lead the HLE-Verified benchmark (54.9%) and perform strongly on DeepSWE v1.1, with Flash Cyber scoring 86.2% on Cybergym (beating GPT-5.5 Cyber) and 47.2% Pass@1 on CWE-Bench. Pricing remains $0.75 per million input tokens and $3.75 per million output tokens through 2026, then doubles in 2027, undercutting rivals like OpenAI and Anthropic. Cyber access is restricted via the Fairwind Program to select partners (e.g., CrowdStrike, Snowflake, government, critical infrastructure). Models are available across Google platforms, with MrBeast promoting them.
Gemini 3.1 Flash-Lite Arrives: Faster, Cheaper
Google unveils Gemini 3.1 Flash-Lite, a cost-efficient variant designed for speed and enterprise use. The model is described as 2.5 times faster than Gemini 2.5 Flash and offers lower costs, with pricing of 0.25 USD per million input tokens and 1.50 USD per million output tokens. It features dynamic Thinking Levels that let users tune the model's reasoning depth. Gemini 3.1 Flash-Lite is available now as a Preview in the Gemini API via Google AI Studio and to enterprises on Vertex AI. Google also notes a 45% improvement in output tempo. In benchmarks, it achieved around 86.9% on the GPQA Diamond test. Google showcases deployment scenarios ranging from translations and content moderation to dashboards and CRM processes, including a Retail Business Agent that can plan and execute multi-step tasks like reporting and dashboard automation.
Gemini 3 Flash Becomes Default AI Mode Model
Google announces Gemini 3 Flash as the default model for its Gemini App, AI Mode in Google Search, and related AI workflows, emphasizing speed and efficiency. The model introduces features such as Agent CC for Gmail and a Disco Browser, and is positioned as faster and more token-efficient than Gemini 2.5. In benchmarks, Gemini 3 Flash reportedly outs as fast or faster than competing models, with a 33.7% score on Humanity’s Last Exam and improved output latency. The system is described as using about 30% fewer tokens on average than Gemini 2.5 and being three times faster, with costs cited at roughly $0.50 per million input tokens and $3 per million output tokens. Access for enterprises is via Vertex AI and Gemini Enterprise, while developers can use the Gemini API in Google AI Studio, Gemini CLI, and the new Google Antigravity platform. The rollout is described as global, establishing Gemini 3 Flash as a foundational AI capability for search and apps.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.