AI Research

Start building with Gemini 2.5 Flash

Published

5 months ago

April 17, 2025

Today we are rolling out an early version of Gemini 2.5 Flash in preview through the Gemini API via Google AI Studio and Vertex AI. Building upon the popular foundation of 2.0 Flash, this new version delivers a major upgrade in reasoning capabilities, while still prioritizing speed and cost. Gemini 2.5 Flash is our first fully hybrid reasoning model, giving developers the ability to turn thinking on or off. The model also allows developers to set thinking budgets to find the right tradeoff between quality, cost, and latency. Even with thinking off, developers can maintain the fast speeds of 2.0 Flash, and improve performance.

Our Gemini 2.5 models are thinking models, capable of reasoning through their thoughts before responding. Instead of immediately generating an output, the model can perform a “thinking” process to better understand the prompt, break down complex tasks, and plan a response. On complex tasks that require multiple steps of reasoning (like solving math problems or analyzing research questions), the thinking process allows the model to arrive at more accurate and comprehensive answers. In fact, Gemini 2.5 Flash performs strongly on Hard Prompts in LMArena, second only to 2.5 Pro.

2.5 Flash has comparable metrics to other leading models for a fraction of the cost and size.

Our most cost-efficient thinking model

2.5 Flash continues to lead as the model with the best price-to-performance ratio.

A graph showing Gemini 2.5 Flash price-to-performance comparison

Gemini 2.5 Flash adds another model to Google’s pareto frontier of cost to quality.*

Fine-grained controls to manage thinking

We know that different use cases have different tradeoffs in quality, cost, and latency. To give developers flexibility, we’ve enabled setting a thinking budget that offers fine-grained control over the maximum number of tokens a model can generate while thinking. A higher budget allows the model to reason further to improve quality. Importantly, though, the budget sets a cap on how much 2.5 Flash can think, but the model does not use the full budget if the prompt does not require it.

Plot graphs show improvements in reasoning quality as thinking budget increases

Improvements in reasoning quality as thinking budget increases.

The model is trained to know how long to think for a given prompt, and therefore automatically decides how much to think based on the perceived task complexity.

If you want to keep the lowest cost and latency while still improving performance over 2.0 Flash, set the thinking budget to 0. You can also choose to set a specific token budget for the thinking phase using a parameter in the API or the slider in Google AI Studio and in Vertex AI. The budget can range from 0 to 24576 tokens for 2.5 Flash.

The following prompts demonstrate how much reasoning may be used in the 2.5 Flash’s default mode.

Prompts requiring low reasoning:

Example 1: “Thank you” in Spanish

Example 2: How many provinces does Canada have?

Prompts requiring medium reasoning:

Example 1: You roll two dice. What’s the probability they add up to 7?

Example 2: My gym has pickup hours for basketball between 9-3pm on MWF and between 2-8pm on Tuesday and Saturday. If I work 9-6pm 5 days a week and want to play 5 hours of basketball on weekdays, create a schedule for me to make it all work.

Prompts requiring high reasoning:

Example 1: A cantilever beam of length L=3m has a rectangular cross-section (width b=0.1m, height h=0.2m) and is made of steel (E=200 GPa). It is subjected to a uniformly distributed load w=5 kN/m along its entire length and a point load P=10 kN at its free end. Calculate the maximum bending stress (σ_max).

Example 2: Write a function evaluate_cells(cells: Dict[str, str]) -> Dict[str, float] that computes the values of spreadsheet cells.

Each cell contains:

Or a formula like "=A1 + B1 * 2" using +, -, *,/ and other cells.

Requirements:

Resolve dependencies between cells.

Handle operator precedence (*/ before +-).

Detect cycles and raise ValueError("Cycle detected at ").

No eval(). Use only built-in libraries.

Start building with Gemini 2.5 Flash today

Gemini 2.5 Flash with thinking capabilities is now available in preview via the Gemini API in Google AI Studio and in Vertex AI, and in a dedicated dropdown in the Gemini app. We encourage you to experiment with the thinking_budget parameter and explore how controllable reasoning can help you solve more complex problems.

from google import genai

client = genai.Client(api_key="GEMINI_API_KEY")

response = client.models.generate_content(
  model="gemini-2.5-flash-preview-04-17",
  contents="You roll two dice. What’s the probability they add up to 7?",
  config=genai.types.GenerateContentConfig(
    thinking_config=genai.types.ThinkingConfig(
      thinking_budget=1024
    )
  )
)

print(response.text)

Python

Find detailed API references and thinking guides in our developer docs or get started with code examples from the Gemini Cookbook.

We will continue to improve Gemini 2.5 Flash, with more coming soon, before we make it generally available for full production use.

^*_{^{Model pricing is sourced from Artificial Analysis & Company Documentation}}

Source link

AI Research

Mira Murati’s Thinking Machines Lab Publishes First Research on Deterministic AI Models

Published

49 minutes ago

September 12, 2025

James Dargan

Thinking Machines Lab, the AI research company founded by former OpenAI CTO Mira Murati, has released its first public research under a new blog series titled Connectionism. Backed by $2 billion in seed funding and a team of former OpenAI researchers, the lab is focused on solving fundamental challenges in AI.

The inaugural post, authored by Horace He, explores how randomness in large language model inference arises from GPU kernel orchestration. The research outlines techniques to create deterministic responses, a breakthrough with potential applications in enterprise reliability, scientific research, and reinforcement learning. The publication marks a rare glimpse into one of Silicon Valley’s most closely watched AI startups as it prepares its first product launch.

Source link

AI Research

When you call Donatos, you might be talking to AI

Published

1 hour ago

September 12, 2025

WSYX Staff

If you call Donatos Pizza to place an order, you might be speaking with artificial intelligence.

The Columbus-based pizza chain announced that it has completed a systemwide rollout of voice-ordering technology powered by Revmo AI. The company says the system is now live at all 174 Donatos locations and has already handled more than 301,000 calls since June.

Donatos Reports Higher Order Accuracy, More Efficient Operations

According to Donatos, the AI system has converted 71% of calls into orders, up from 58% before the rollout, and has achieved 99.9% order accuracy. The company also says the switch freed up nearly 5,000 hours of staff time in August alone, allowing employees to focus more on preparing food and serving in-store customers.

“Our focus was simple: deliver a better guest experience on the phone and increase order conversions,” Kevin King, President of Donatos Pizza, said in a statement.

Ben Smith, Donatos’ Director of Operations Development, said the change provided immediate relief on the phones, allowing staff to redirect time to order accuracy and hospitality.

Donatos said it plans to expand the system to handle more types of calls and to make greater use of its centralized answering center. The company did not say whether it plans to reduce call center staffing or rely more heavily on automation in the future.

Other chains report trouble with AI ordering systems

Taco Bell recently started re-evaluating its used of AI to take orders in the drive-thru after viral videos exposed its flaws. In one well-known video, a man crashed the system by ordering 18,000 cups of water. The company is now looking at how AI can help during busy times and when it’s appropriate for a human employee to step in and take the order.

Last year, McDonald’s ended its AI test in 100 restaurants after similar problems surfaced. In one case, AI added bacon to a customer’s ice cream. A McDonald’s executive told the BBC that artificial intelligence will still be part of the chain’s future.

Source link

AI Research

Why Ibex Stock Surged 41% to All-Time Highs Today (Hint: It’s Artificial Intelligence)

Published

1 hour ago

September 12, 2025

The Motley Fool

Key Points

Ibex reported record revenue for its fourth quarter and full year of 2025.

Ibex is expanding its AI tools and targeting new verticals.

The stock hit an all-time high on Sept. 12.

10 stocks we like better than Ibex ›

Shares of little-known company Ibex (NASDAQ: IBEX) went parabolic today, shooting 41.1% higher in early-morning trading. The stock was still trading around 33% up at 1:15 p.m. ET Friday.

Ibex is a business process outsourcing company, providing a wide array of services such as customer and technical support, lead generation, surveys, and business intelligence and analytics.

Where to invest $1,000 right now? Our analyst team just revealed what they believe are the 10 best stocks to buy right now. Continue »

Turns out, Ibex’s efforts to build a digital business have already started to pay off, and that is drawing attention to the stock today. The keyword here is artificial intelligence (AI).

Image source: Getty Images.

AI-driven growth

Ibex reported numbers for its 2025 fourth quarter and fiscal year (ended June 30) after the Sept. 11 market close. Ibex’s Q4 revenue jumped 18% year over year to $147 million, driven by strong growth in its top three markets: retail and e-commerce; healthcare; and travel, transportation, and logistics.

The real deal, however, is what Ibex’s full earnings report looked like:

Record fourth-quarter and full-year revenue
Highest revenue growth in 11 quarters
Fastest revenue growth in three years for the full year
Record free cash flow

These are big milestones, but they’re not really why Ibex stock is going to the moon. It’s these words from CEO Bob Dechant: “Importantly, this quarter marked the shift from proof of concept for our AI solutions to full-scale deployments, setting the table for future growth.”

Ibex is “transforming into a digital-first business” by leveraging AI through its Wave iX platform, which uses generative AI to improve customer experiences. Earlier this month, Ibex said it is targeting the government sector now.

What’s next for Ibex stock?

The company’s capital expenditures more than doubled to $18.4 million in 2025, driven by capacity expansion. Ibex generated record free cash flow of $27.3 million in the year and repurchased nearly 3.9 million shares, almost 23% of its outstanding shares.

Following Ibex’s strong earnings report, analysts at RBC Capital were quick to raise their price target on the stock to $39 per share from $31 a share. Ibex stock already hit an all-time high of $42.99 per share today.

With Ibex projecting 7.5% revenue growth at the midpoint for FY 2026 and capital expenditure of $20 million to $25 million on further expansions, this is one stock you should have on your radar.

Should you invest $1,000 in Ibex right now?

Before you buy stock in Ibex, consider this:

The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Ibex wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.

Consider when Netflix made this list on December 17, 2004… if you invested $1,000 at the time of our recommendation, you’d have $649,037!* Or when Nvidia made this list on April 15, 2005… if you invested $1,000 at the time of our recommendation, you’d have $1,086,028!*

Now, it’s worth noting Stock Advisor’s total average return is 1,056% — a market-crushing outperformance compared to 188% for the S&P 500. Don’t miss out on the latest top 10 list, available when you join Stock Advisor.

See the 10 stocks »

*Stock Advisor returns as of September 8, 2025

Neha Chamaria has no position in any of the stocks mentioned. The Motley Fool has no position in any of the stocks mentioned. The Motley Fool has a disclosure policy.

The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.

Source link