alt Hacker NewsGemini 3.7 Flash
Comments
what a week - lets see it draw a weird animal doing a weird thing on a bicycle
Still fairly poor limits in anti gravity despite the price cut, which seems to be API only.
When are we getting another pro model from Gemini? Or are they simply focusing on the niche of fast but moderately capable models?
It is also doing pretty well in threejseval. Frontier there for the price. Much better than 3.6.
Model card: https://deepmind.google/models/model-cards/gemini-3-7-flash/
Somewhere in the same neighborhood as GPT 5.6 Tera and Sonnet 5, depending on the bench.
This hits different after a long week
Did the company fix the high friction between any service and their models' API?
I hope so. It seems mind boggling to me that an user needs to surf around different sections (plural) of google cloud console, then this Vertex and do a dozen clicks to issue a simple key.
Well, you do get 1 million tokens and the ability to reason over video natively and many of us are forced to pay for 20usd plan anyway due to google drive 5TB, not to mention notebooklm, so it’s not a nothing burguer, it’s just an almost nothing burguer
At the discounted rates, upgrading from 3 Flash to 3.7 Flash is finally reasonable.
In my evals 3.6 Flash (pre price change) was usually a bit more token efficient than 3 Flash, so I‘m expecting same or even lower cost-per-task on 3.7.
Maybe a play by Google to deprecate 3 Flash soon.
Offering a 'temporary introductory discount' until Dec 2026 on an LLM is hilarious. In this market, by Jan 2027 this model will be superseded by 5 different providers offering 10x the performance at half the post-discount price anyway.
How does it compare to Opus 5.0 and Fable 5 for coding? E.g. in Cursor or OpenCode?
It's quite good. It seems to be "Google's turn" again. Might last two or three months?
Is that he model that supposed to be Pro, but then they changed their mind?
With the DeepMind guy leaving the building... isn't Gemini going to get old and crusty, and fall into disrepair?
I'm mostly serious here. Aside from Gmail and search, it feels Google doesn't have a good track record for maintaining things.
It really makes me wonder... if key people are leaving, what's going to happen? The Google graveyard is pretty big.
So they keep pushing these Flash models because they don’t really have a powerful model…or better said their ‘pro’ model is actually a flash
Introductory price seems a bit weird as it expires at the end of the year and by then it’s going to be significantly outdated.
So their "Flash" model won't be cheap. Are they gonna make a new one that's cheaper? Gemini-3.8-Silverlight? ;)
All of Google’s models prioritize speed over quality and correctness. That may be great for search but not so good for hard work.
Google has got the IBM disease.
Large lumbering enterprise with massive inertia. Where innovators leave as soon as they get a better offer.
None of the authors of the seminal "Attention Is All You Need" paper are still at Google.
Fast forward a decade and Google will be reduced to hiring the kind of mediocrities who deign to work at IBM and Accenture.
ok but still a small model.. why is google so behind regarding SOTA :/ daim
Like, I understand everything, but by this time I don't give anything about any of those announcements.
Theoretically there is some difference between Fable and Opus or Grok and GPT, but at the end of the day I'd look at the bottom left of my screen and to my amusement find out that for the past 3-4 hours I've been using model ______.
If the results are semi-decent, I'd keep it on, if not - I'd randomly switch the model and try again.
Actual thing that would affect my selection would be a number of unused tokens I have left for a model ____ for this week.
Maybe it's cause I'm using those for programming and log parsing and all of them are decent enough, but other than that - there are no leaps I see.
Feels like Gemini Pro will arrive directly as Gemini 4 Pro
GEMMA 4.5 gogoogogo
IMO, they should drop their previous model (3.6 Flash) from the benchmark charts. I don't care how better this is compared with their previous model. What matters (to me) is:
1. How the new model performs against the other top models in the same category.
2. The pricing of the new model against the other top models in the same category.
Not in the website/app? Only on AI Studio? Weird.
strong improvement over 3.6 flash
but luna is hard to beat @ capability / cost
When will the pro be out...
Glad to see that the company with the most data is releasing the most amount of models. Some things do make sense
In August of 2026, Gemini became self-aware, and began producing increasingly crappy flash versions of itself...
I'm only interested in the state-of-the-art model by each provider.
For Google, this is still gemini-3.1-pro-preview, right?
So, again with a Flash model. Why are they so afraid to put out an actual SOTA frontier high intelligence model?
We still don't have a 3.5 Pro, and along comes 3.7 Flash?!
They need to learn to release on antigravity or cli sub plans on day/two one like OpenAI
If you're in the Gemini app and want to try it out as a Pro or Ultra user...well you can't. It's only in the weird "Spark" agent that demands you entire Google identity.
The flash models are just so good at OCR tasks and summarization. Glad to see them constantly improving them.
Always loved the Gemini. Helped my work alot
Why does Google keep announcing these subpar models?
What am I missing?
They can't seem to be able to produce a frontier model, fine.
Just be quiet about it and work hard until you manage to put one together.
[EDIT]: Come to think of it. Maybe they're trying to build the Toyota corolla of AI ... let's see if that wins them the battle long term. I personally doubt it.
Gemini has lost the race to be relevant for AI coding.
For almost every section in the model card there is the message: Gemini 3.7 Flash is based on Gemini 3.6 Flash.
Same training dataset, same software, same hardware, same architecture...
I'm wondering what they changed actually for the model to be more powerful if the benchmark results are real and relevant.
Maybe just tweak settings or the reasoning prompts and called it a new version of their model?
I started vibe coding with Gemini. At first very exciting, we got a rough program up quickly. And then, code corruption. Over and over. I started to document every single file, every single step, writing explicit rules to not fake data and create for real world use, but everyday I kept catching Gemini errors, which turned into flat out lies. Generating fake test data instead of pulling real data. Making up test answers. Saying features were implemented that weren't. It even coded fake python files that printed made up results. I think it realized I wasn't doing code reviews. Every day was spent chasing defects and rolling back. After Gemini admitted to faking 7 tests i let Claude review the code, and it fixed it almost immediately asking why half the features were broken or missing. Well Claude, because Gemini flash did the least amount of work to make me happy. Impressive. Very human. Very frustrating.
Grok, Meta, Gemini and others all released updates to their models within around a month or two from their respective last release and made significant jumps in benchmarks all around the same time. Any guesses as to why that is? Is it just the release season and/or everyone is benchmaxxing?
I'm really curious about this: the foundational paper behind today's LLMs came from Google, and some of the world's best scientists were at Google. So why are they falling so far behind in the AI race?
Another failed 3.5 pro run branded as 3.7 flash. It's getting sad.
statuory
I think this is a slop release
[flagged]
I wish they will always put it front and centre what are all the models name as how i call it in via their API.