model release

Google releases Gemini 3.5 Flash at half the price of frontier models, announces Omni world model

TL;DR

Google released Gemini 3.5 Flash, priced at half to one-third the cost of comparable frontier models, and announced it will become the default model in the Gemini app globally. The company also unveiled Omni, a world model for simulating physical environments, and Gemini Spark, an AI agent in beta testing.

2 min read
0

Gemini 3.5 Flash at a glance

Context window1049K tokens
Input$1.5/1M tokens
Output$9/1M tokens

How Gemini 3.5 Flash ranks now

BenchmarkScoreRankSource
Arena Elo1478 Elo26 of 141LMArena, Nov 2026 ↗
GPQA92.2%25 of 164Artificial Analysis, Jul 2026 ↗
Speed (tok/s)167 tok/s17 of 103Artificial Analysis, Nov 2026 ↗

Rank is against every model in our database with a sourced score on that benchmark, each counted once with its best score.

All Gemini 3.5 Flash specs and scores →

Google releases Gemini 3.5 Flash at half the price of frontier models, announces Omni world model

Google released Gemini 3.5 Flash at its I/O developer conference on May 19, 2026, positioning the model at half to one-third the price of comparable frontier models according to CEO Sundar Pichai. The model will immediately become the default in the Gemini app and AI mode in search globally.

Pricing and performance

Google claims Gemini 3.5 Flash delivers "cutting-edge capabilities" at 50-67% lower cost than competing frontier models. Specific pricing per million tokens was not disclosed. Pichai described the model as "remarkably fast" in a pre-event briefing, with Google stating users "no longer have to trade quality for latency."

The company said it strengthened cybersecurity defenses to make 3.5 Flash "less likely to generate harmful content and mistakenly refuse to answer safe queries." No benchmark scores were provided.

Gemini 3.5 Pro delayed

Gemini 3.5 Pro, the heavier-weight version, is currently in internal use but won't launch publicly until June 2026. Google did not provide technical specifications or performance comparisons for the Pro version.

Omni world model for physical simulation

Google announced Omni, a world model designed to simulate physical environments and predict outcomes based on user actions. The model supports image and audio input and will be integrated into Flash, the Gemini App, Google Flow, and YouTube Shorts.

According to Google, Omni can "edit the action, add in new characters or objects" in user-uploaded videos. World models are primarily used in robotics and gaming applications. DeepMind has researched world models extensively, though Google did not specify how much of that research informed Omni's development.

Gemini Spark agent in beta

Google introduced Gemini Spark, an AI agent that can "reason across information in connected apps" and take actions on behalf of users. The agent launches in beta next week for trusted testers and Google AI Ultra subscribers only. No timeline for general availability was provided.

The agent represents Google's push into agentic AI as the company attempts to create deeper integrations across its product suite.

Market context

The announcements come as OpenAI and Anthropic, both reportedly preparing for IPOs in 2026, dominate AI market attention. Anthropic recently released its Mythos model, which the company claims discovered thousands of previously unknown software vulnerabilities.

Google's capital expenditure on AI infrastructure has increased significantly, with Wall Street expecting the company to demonstrate return on investment through product integrations.

What this means

Google's aggressive pricing on Gemini 3.5 Flash directly challenges OpenAI and Anthropic on cost, a critical factor for developers building at scale. The lack of disclosed benchmarks makes it impossible to verify performance claims against GPT-4 or Claude 3.5 Sonnet. Omni's world model capabilities could differentiate Google in robotics and video editing applications, but commercial viability depends on accuracy and reliability metrics not yet public. The staggered rollout of Spark and delayed Pro model suggest Google is still validating core capabilities before broad deployment.

Source: cnbc.com ↗

Related Articles

model release

Google releases Genie 4 to AI Ultra subscribers with real-time prompt editing and native audio

Google is rolling out Genie 4, which it calls its most realistic and interactive world model yet, to eligible Google AI Ultra subscribers in Project Genie. The update adds real-time native audio, mid-session text prompting, a world-building agent and a community gallery. Genie 3 also gains sessions of up to 3 minutes.

model release

OpenAI releases GPT-6.2 Astra with 1.1M context at $8 input, $40 output per 1M tokens

OpenAI released GPT-6.2 Astra on Nov 17, 2026, listing a 1.1M-token context window and pricing of $8 input and $40 output per 1M tokens. OpenAI reports new highs on its evals at less than a third of the cost per task of GPT-6 Astra.

model release

RadixArk releases DaisyRose-27B, lifting Terminal-Bench 4.1 pass@1 from 5.6% to 12.0%

RadixArk released DaisyRose-27B, its first model, built on Qwen3.8-27B and trained with reinforcement learning on 1,200 curated agent tasks. According to RadixArk, it raises Terminal Bench 4.1 pass@1 from 5.6% to 12.0% and scores higher than the base model on all seven reported benchmarks. Weights ship under Apache 2.0 in BF16, FP8 and NVFP4.

model release

Microsoft releases Aion 1.0 Instruct, a 2.3B on-device SLM with 32K context under MIT license

Microsoft's Windows Applied Sciences Group has published Aion 1.0 Instruct, a roughly 2.29B-parameter instruction-tuned small language model with a 32,768-token context window under the MIT license. A further optimized version is slated to become the Windows OS SLM by the end of January 2027.

Comments

Loading...