Vai al contenuto

Lingua interfaccia: IT. Contenuti editoriali: inglese. Prezzi riferiti a Stati Uniti.

Sir CompareConfronti con le proveYouTubeTelegram
Stati Uniti · USD

Notizia

Gemini 4 Hydrogen: what changes for free users and API pricing

Aggiornato

Foto illustrativa: AI Tools & SaaS: Code on computer monitor (Unsplash)
Foto illustrativa

Hydrogen takes Flash's place across Google products

Google has replaced Gemini 3.8 Flash with a new model called Gemini 4 Hydrogen across its products, according to Ars Technica. The name follows Google's elemental scheme, but the job is the one Flash used to do: the faster, cheaper option most people in the ecosystem touch.

The rollout is looser than the recent Argon and Carbon releases. Those started with cybersecurity researchers and, even at general availability, stayed behind the priciest AI subscription. Hydrogen access begins immediately, and the limits are described as much more relaxed.

Google says the model is a major step up from the previous Flash. It runs at one of four thinking levels: low, medium, high, and a new xhigh setting.

I prodotti di cui parliamo

Speed, fewer fabricated claims, better code scores

The numbers come from Google's own analysis. In high mode, Hydrogen is 2.5 times faster than Flash 3.8. It also cuts fabricated claims by more than five times, again per Google's figures.

Code generation gets a lift too: Hydrogen more than triples 3.8 Flash's score on Terminal-Bench. In some cases it even trades blows with Argon, the bigger model. With xhigh mode it scored 54 on the Artificial Analysis Intelligence Index, against 50 for Argon.

Google frames Hydrogen as a key piece of agentic coding work. Argon and Carbon stay positioned as the best for long-horizon tasks and security. Hydrogen's speed suits simpler code and test runs without burning through token budgets.

Search gets the model, the Gemini app partly doesn't

Hydrogen is also being wired into Google Search. The company calls it the biggest update yet to its AI-powered search tools. The model should handle long, multipart questions better, and answers in AI Overviews and AI Mode will carry more visual elements.

Google also says Hydrogen's fan-out search has been tuned to link to more sources. Ars Technica notes it has heard that claim from Google before.

Access in search is reportedly open to everyone. The Gemini app and AI Studio are tighter: AI Pro and Ultra subscribers get Hydrogen now, while AI Plus and free users stay on the older 3.5 Flash-Lite.

API price stays flat, Argon and Carbon go up

Developers reach Hydrogen through the API at $1.50 per million input tokens and $7.50 per million output tokens. That is unchanged from 3.8 Flash, so moving to the new model does not raise the per-token bill.

The bigger models are a different story. Carbon and Argon currently cost $2 per million input and $10 per million output. Starting next month those prices rise to $4 and $20.

If you build on the cheaper tier, the swap to Hydrogen keeps your costs where they are. If your workload leans on Argon or Carbon, the next invoice will look different.

What to check before you switch

The performance figures are Google's, not independent measurements. I haven't run Hydrogen myself, so treat the 2.5x speed claim and the Terminal-Bench jump as vendor numbers until outside tests appear.

The practical split is simple. On the API, the price is the same as Flash, so the switch costs nothing per token. In the Gemini app, free and AI Plus users keep the older model for now.

One thing worth watching: Google's promise of more linked sources in search. Ars Technica flags that the company has made similar claims before, so the fan-out tuning is worth checking against real queries.

Related coverage

Source: Ars Technica · AI

Hardware per provare l'AI in locale

Link affiliati: possiamo ricevere una commissione sugli acquisti idonei, e non cambia i nostri dati né le classifiche. Aprono i risultati di ricerca di Amazon: prezzi e disponibilità si vedono lì.

Video

Cosa è cambiato

  • Pagina aggiornata

Come metto in fila i prodotti

La posta di Sir Compare

Novità su Strumenti AI e SaaS? Te le scrivo io.

Scegli cosa ti interessa: ti scrivo solo quando esce qualcosa di nuovo — una guida, una notizia, un prezzo verificato che scende.

Passo 1 di 2

Cosa ti interessa?

Scegline quante vuoi. Se non scegli niente, ricevi le novità di tutte le sezioni.

Cosa vuoi ricevere
  • Solo quello che scegli
  • Mai email vuote
  • Esci con un clic

Il tuo assistente

Apri la pagina completa →