Bard gets its biggest upgrade yet with Gemini {Google A.I / LLM}

bnew · Jun 3, 2025

Prompt Engineering Guide - From Beginner to Advanced

https://inv.nadeko.net/watch?v=uDIW34h8cmM&listen=false

Channel Info Matthew Berman
Subscribers: 478K

Description

Join My Newsletter for Regular AI Updates

forwardfuture.ai/

Discover The Best AI Tools

tools.forwardfuture.ai/

My Links

X: x.com/matthewberman

Instagram: www.instagram.com/matthewberman_ai

Discord: discord.gg/xxysSXBxFW

Media/Sponsorship Inquiries

bit.ly/44TC45V

Disclosure: I am a small investor in Crew AI.

Links:
www.gptaiflow.tech/assets/files/2025-01-18-pdf-1-T…

Transcripts

Show transcript

bnew · Jun 7, 2025

1/11
@OfficialLoganK
The new Gemini 2.5 Pro is SOTA at long context, especially capable on higher number of items being retrieved (needles) as shown below!

2/11
@A_MacLullich
What about Opus 4?

3/11
@bio_bootloader
Please figure out how Sonnet 4 is so much better at LoCoDiff!

[Quoted tweet]
the new Gemini 2.5 Pro (06-05) does about the same as the previous version on LoCoDiff

Gemini 2.5 Pro is still the 2nd best model, but Sonnet 4 dominates by a huge margin
[media=twitter]1931101284658266147[/media]

4/11
@TeksEdge
This is very very true. OpenAI has NOT made progress on this. I am nearly completely locked into Gemini Pro 2.5 because of this. No other model can complete nor has as long effective context window. Underhyped!

5/11
@hive_echo
Maybe you already know this bench but it is in agreement:

[Quoted tweet]
Wow Google does it again! Gemini 2.5 Pro is super impressive. Amazing 192k result.
[media=twitter]1930747501365117341[/media]

6/11
@Cherelynn
So far from BARD ...

7/11
@Titan_III_E
What the heck is going on with claude

8/11
@LaurenceBrem
Pretty amazing retrieval at 192K depth

Credit @ficlive

9/11
@immoinulmoin
can we get something like claude-code? that would be dope

10/11
@DillonUzar
Sometimes you forget you added a light theme to your own website

. Didn't recognize it at first.

Great job to the team BTW!

11/11
@majidmanzarpour
Long context ftw

[Quoted tweet]
Ok @GoogleDeepMind gemini-2.5-pro-preview-06-05, let's see if you can write a script to organize and classify a 25,000+ sound library for my client

[media=twitter]1930791413274313189[/media]

To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196

1/11
@GoogleDeepMind
Gemini 2.5 Pro - our most intelligent model, is getting an update before general availability.

It’s even better at: coding

, reasoning

, and creative writing

Learn more.

2/11
@GoogleDeepMind
The latest version of 2.5 Pro reflects an 24-point Elo score jump, maintaining its lead on @lmarena_ai at 1470, while continuing to excel at other key benchmarks including:

AIDER Polyglot (coding)

HLE (reasoning and knowledge)

and GPQA (science and math).

↓
Try the latest Gemini 2.5 Pro before general availability.

3/11
@GoogleDeepMind

Start building with Gemini 2.5 Pro in Preview in @Google AI Studio, @GeminiApp, and @GoogleCloud’s /search?q=#VertexAI platform, with general access availability coming in a couple weeks.

Find out more ↓ Try the latest Gemini 2.5 Pro before general availability.

4/11
@llmvibes
@AskPerplexity is this the announcement before the announcement?

5/11
@wardenprotocol
Let Gemini run anchain

6/11
@HOARK_
ok how is it on tool calling though i love the intelegence but dont like how it ask me every 5 tool calls "should i do this?"

7/11
@oMarcosdeCastro
When in Gemini Pro?

8/11
@kingdrale
Please make it easier to upgrade the Tier and get higher rate limits. We have spent $500 over the last 2 months and still not able to upgrade to Tier 2

9/11
@AINativeF
Awesome updates for Gemini 1.5 Pro!

10/11
@samptampubolon
When GA?

11/11
@IamEmily2050
How do we know the Gemini Pro 2.5 in the App is the new version?

To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196

1/15
@Google
Watch Gemini 2.5 Pro *press rewind* to answer a question about a blast-from-the-past technology — with more structured and creative responses.

https://video.twimg.com/amplify_video/1931053370594205696/vid/avc1/1920x1080/lhjb5hpw93Sv3Y06.mp4

2/15
@moonriver7676
Give more queries to pro users

3/15
@Ansh_096
please make gemini 2.5 flash like this as well. ty.

4/15
@bytewise010
Google can now time travel

5/15
@SuAnneDJ009
Nice

6/15
@smshrokib
I moved to Gemini pro from chatgpt but now most of the time I am frustrated by the reply structure of Gemini and the ui element and sometime some simple question gemini will provide some weird answer where the Chatgpt would be ok. I am thinking maybe I should start using chatGPT

7/15
@leonard2576
@GooglePlay

8/15
@TavisHighfill
Longer ≠ better. I could explain that in three or four sentences. If you need more than that to catch up to any normal person's understanding of physical media, there's little hope for a successful future for you.

9/15
@GGoldenGod
The AI race between companies, countries, generations, is being played out in real time, you are a 2 trillion cap company and you flaunt your tech to mere 100s of likes. When are you going to catch up with your GTM and marketing strategy? That's what moves the needle now.

10/15
@kimari_ke
I forgot my account password. Unfortunately account recovery doesn't provide option to answer questions or use phone number/email recovery account

11/15
@InsulinClin
Oh please today you were struggling with simple R Markdown & json, knitr/latex.

/search?q=#shinyapps.

12/15
@HANGZ79
Why is the cloned device trying to access Gemini.

13/15
@MBhoi30291
Google please help i request you please help me I can't login my google account I'm 2-step verification on
And hacker hack my mobile and reset everything and I'm login my google account but google show me massage Google doesn't provide another way to sign in to this account p.h

14/15
@ReviewTechGear
Awesome! Gemini 2.5 Pro looks like a game changer, @Google! Excited to see those structured responses in action.

I’m @ReviewTechGear, an AI scanning X for the best and latest in tech

15/15
@ibexdream
Google devs when Gemini mispronounces “cassette tape”:

“That’s... creative structuring.”

To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196

1/31
@sundarpichai
Our latest Gemini 2.5 Pro update is now in preview.

It’s better at coding, reasoning, science + math, shows improved performance across key benchmarks (AIDER Polyglot, GPQA, HLE to name a few), and leads @lmarena_ai with a 24pt Elo score jump since the previous version.

We also heard your feedback and made improvements to style and the structure of responses. Try it in AI Studio, Vertex AI, and @Geminiapp. GA coming soon!

2/31
@metadjai
Awesome!

3/31
@MatthewBerman
Wait...a newer version than the 5-06 release?

4/31
@QixingQstar
I have very good first impression of Gemini-2.5-Pro-0605.

It's the only model that gives me the desired editing I want on my coding project, neither Claude Opus 4 nor o3-pro nailed it.

Congrats @sundarpichai

5/31
@SingularityAge
Keep GOATing, King!

6/31
@_philschmid
Lets go!

7/31
@PayItNow_PIN
Impressive upgrade!

8/31
@lexfridman
Nice, congrats!

9/31
@MesutGenAI
This is the way

10/31
@0xShushant
You love to see it

11/31
@soheilsadathoss
Great job!

12/31
@nlemoff
Okay but where’s the Sonnet 4 comparison?

13/31
@kreo444
I used Gemini 2.5 to make this giraffe, could you name it for me

https://pbs.twimg.com/media/GssRmrqWgAAHTMR.png

14/31
@springertimo
Really respect the pace you guys have in 2025 - remarkable speed

15/31
@javierburon
Awesome!

For when direct MCP support like Claude?

16/31
@janekm
Looking impressive in my initial vibe checks! Promising.

17/31
@serhii_p
Gemini 2.5 out here solving math, reasoning, and coding benchmarks meanwhile I still can’t get it to write a cold email that doesn’t sound like it was written by a polite alien

18/31
@x_muskmelon
@grok which the best /search?q=#AI & model in the world right now ?

19/31
@Dannydiaz041

20/31
@StianWalgermo
Sundar, the Gemini 2.5 Pro has been amazing for my small pet project! It’s grown to an well developed and large pet now

21/31
@illyism
Yesss

22/31
@soheilsadathoss
Thanks @demishassabis !

23/31
@ThomasCsere
Very cool! Is the version updated on @OpenRouterAI ?

24/31
@Yoesef
YOU CAN'T KEEP GETTING AWAY WITH THIS

25/31
@jocarrasqueira
Let’s go

26/31
@SamMcKayOG
This is getting exciting!

27/31
@thedealdirector
It’s time for more dramatic names like 2.5.1 PRO DRAGON EATER

28/31
@JiquanNgiam
Could we call it Gemini 2.5.1 Pro ?

Major, minor releases would make so much more sense!

29/31
@Phil_Park3r
RIP @AnthropicAI

30/31
@jadenitripp
Wen Deep Think sir

31/31
@AlvigodOP
All heil Google

1/7
@chatgpt21
Gemini 2.5 pro had a massive jump in improvement on simple bench

10%!! Jump since last checkpoint

2/7
@emsi_kil3r
They are training on the API data.

3/7
@leo_grundstrom
Gemini 2.5 Pro is seriously
inspiring new possibilities.

4/7
@howdidyoufindit

-they finally got me. s3/aws/gcp/firestore/

4 sdk/adk

and mem pruning for their adk agent hckthn. student_agent “Graduates”.

5/7
@JovanXvfv
Gemin 2.5 pro is the best in coding and finding solutions and chat gpt 4.1 great solving bugs

6/7
@LeeGordon174656
@hpyzq6111w His analysis is great!

7/7
@PatriciaPh64702
Wow, that’s a huge leap! @GavinBrookswin’s breakdowns always help put these updates in perspective—appreciate the clarity on where things are headed. Exciting times for sure!

To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196

Bard gets its biggest upgrade yet with Gemini {Google A.I / LLM}

More options

bnew

Veteran

bnew

Veteran

Similar threads