Google ships three new Gemini Flash models as its Pro flagship slips
Google announced Gemini 3.6 Flash, 3.5 Flash-Lite and 3.5 Flash Cyber on July 21, 2026. The new workhorse model cuts output token use by 17% and lowers prices, as the delayed Gemini 3.5 Pro remains absent. Google also confirmed Gemini 4 is in training.
Google added three new members to its Gemini model family on July 21, 2026: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite and the security-focused Gemini 3.5 Flash Cyber. The announcement was posted on the company's official blog under the byline of Tulsee Doshi from the Gemini team. The models rolled out to developers through Google AI Studio and Vertex AI, and Gemini 3.6 Flash also arrived in GitHub Copilot the same day.
Positioned as Google's "workhorse" model, Gemini 3.6 Flash promises gains in coding, knowledge work and multimodal tasks over its predecessor. The company says it uses 17% fewer output tokens than 3.5 Flash and takes fewer reasoning steps and tool calls to complete multi-step workflows. Google puts the effective cost per completed task at roughly 31% lower, rising to about 71% on agentic coding workloads. Pricing is set at $1.50 per 1M input tokens and $7.50 per 1M output tokens, down from $9 for output on the earlier model. The high-throughput 3.5 Flash-Lite is priced at $0.30 per 1M input tokens and $2.50 per 1M output tokens.
The launch comes against the backdrop of a months-long delay to Google's flagship Gemini 3.5 Pro. According to a Bloomberg report citing 10 current and former employees, the company reset and updated its training data in late June to address weak coding performance, but internal results again fell short. Sundar Pichai had told developers in May the model would ship in June, a target that has now passed. Google instead kept its lineup moving with releases in the lighter, high-volume Flash segment.
The new models sharpen an intensifying price war, with the Flash tier serving as the go-to choice for high-volume, cost-sensitive applications. Google also carved out a dedicated security niche with Gemini 3.5 Flash Cyber, designed to detect, validate and patch code vulnerabilities at scale. The company confirmed that training for its next major generation, Gemini 4, is already underway in parallel. The result is a week in which Google signaled it can hold its release cadence — and its place in the race — despite the stumble on the Pro side.
Moonshot AI releases Kimi K3, the largest open-weight model to date
China's Moonshot AI unveiled Kimi K3, a 2.8-trillion-parameter model billed as the largest open-weight system yet. It offers a 1-million-token context and native multimodal input, with full weights due July 27.
xAI's Grok 4.6 rejoins the AI frontier and takes the cost-efficiency lead
SpaceXAI, formerly xAI, released Grok 4.6 on August 12, 2026. The model matches GPT-5.6 Sol at 61 on the Artificial Analysis Intelligence Index while holding prices flat, delivering frontier intelligence at less than half the cost of its rivals.