Google's Gemini 4 Argon costs $2 per million input tokens, until prices double in November
What happened
Google released Gemini 4 Argon on Oct. 13 at an introductory $2 per million input tokens and $10 per million output tokens. The Verge reports that is about a fifth of the per-token price of OpenAI's GPT-6 Astra and Anthropic's Fable 5 models, and that the rates double when the introductory period ends in late November. Google says Argon beats rival frontier models on benchmarks, including #1 on the Vals Index. Those results are Google's own and have not been independently confirmed in the sources reviewed.
What we know
- Per Google, Argon is available to developers through the paid tier of the Gemini API. Google AI Ultra subscribers can use it in Google AI Studio and Antigravity, its agent-first development platform.
- Per Google, AI Pro and Ultra subscribers get Argon in the Gemini app and in AI Mode in Search. Per The Verge, it is unclear when or whether Plus or free users will get it.
- Google says Argon ranks #1 on the Vals Index, which covers finance, coding, legal and tax work. The Verge adds that Google claims it beats GPT-6 Astra, Claude Fable 5.1 and Claude Opus 5.5 on benchmarks in coding, cybersecurity, science, finance and law.
- Google reports that Argon cut in-memory session preparation time by 40% in its open-source Agent Development Kit and sped up cold imports by 21%. It also reports a 2.85x inference speedup for its Gemma 4 E2B model on an A100 GPU, with 100% token parity.
- Per The Verge, the public launch follows a limited release to selected "trusted cyber defenders" for risk testing. It is Google's first major frontier release since Gemini 3 in late 2025.
What we don't know yet
- Whether Argon's benchmark lead holds up in independent testing. All performance figures in the sources come from Google.
- What Argon costs per task. Per The Verge, models use different numbers of tokens for the same work, so a lower per-token price does not guarantee lower bills.
- Whether Argon will reach Plus or free Gemini users, and when.
- The exact post-promotion prices. The Verge says they double at the end of November, but the Google posts reviewed only call the current rates introductory.
Why it matters
Developers choosing a frontier model for coding and agent workloads can now compare Argon at $2 and $10 per million tokens against rivals priced around five times higher. The comparison is only reliable after the November price increase and independent benchmarks.
Claims
- ConfirmedGoogle launched Gemini 4 Argon on Oct. 13, 2026.
- ConfirmedArgon launches at an introductory price of $2 per million input tokens and $10 per million output tokens.
- UnconfirmedArgon's prices double when the introductory period ends in late November.
- UnconfirmedArgon's introductory price is about a fifth of the per-token price of GPT-6 Astra and Anthropic's Fable 5 models.
- UnconfirmedArgon ranks #1 on the Vals Index.
- ConfirmedArgon is available in the Gemini app and AI Mode in Search to Google AI Pro and Ultra subscribers.
- UnconfirmedArgon cut in-memory session preparation time by 40% in long sessions in Google's Agent Development Kit.
- UnconfirmedGoogle used Argon to achieve a 2.85x end-to-end inference speedup on Gemma 4 E2B on an A100 via vLLM.
- UnconfirmedThe public launch followed a limited release to selected trusted cyber defenders.
Related coverage
Why you can trust this story
70%Source map · 2 outlets / 4 articles
- Google (The Keyword)×3
- Article 1blog.googleofficial
- Article 2blog.googleofficial
- Article 3blog.googleofficial
- The Verge
- Article 1theverge.comreport
How this credibility score is calculated
- Source reliability
- Corroboration
- Primary source
- Atom-verified claims
- No contradiction
- Settled
- Claim attribution
- AI disclosure
Atom-verified claims matched cited source text. Weights are fixed and explainable.
Human accountability
- 2 independent origins / 4 sources
- Drafted by the Newsmesis judgment agent (agent-cli); human approval required before publishing.
Verification ledger · 9
- Claim: Google launched Gemini 4 Argon on Oct. 13, 2026.Status: ConfirmedVerification trailVerifiedChecked by verification engine
At least one extracted atom matched cited source full text.
- Claim: Argon launches at an introductory price of $2 per million input tokens and $10 per million output tokens.Status: ConfirmedVerification trailVerifiedChecked by verification engine
At least one extracted atom matched cited source full text.
- Claim: Argon's prices double when the introductory period ends in late November.Status: UnconfirmedVerification trailUnverifiedChecked by verification engine
No atom-level source-text verification is available.
- Claim: Argon's introductory price is about a fifth of the per-token price of GPT-6 Astra and Anthropic's Fable 5 models.Status: UnconfirmedVerification trailUnverifiedChecked by verification engine
No atom-level source-text verification is available.
- Claim: Argon ranks #1 on the Vals Index.Status: UnconfirmedVerification trailUnverifiedChecked by verification engine
No atom-level source-text verification is available.
- Claim: Argon is available in the Gemini app and AI Mode in Search to Google AI Pro and Ultra subscribers.Status: ConfirmedVerification trailVerifiedChecked by verification engine
At least one extracted atom matched cited source full text.
- Claim: Argon cut in-memory session preparation time by 40% in long sessions in Google's Agent Development Kit.Status: UnconfirmedVerification trailUnverifiedChecked by verification engine
No atom-level source-text verification is available.
- Claim: Google used Argon to achieve a 2.85x end-to-end inference speedup on Gemma 4 E2B on an A100 via vLLM.Status: UnconfirmedVerification trailUnverifiedChecked by verification engine
No atom-level source-text verification is available.
- Claim: The public launch followed a limited release to selected trusted cyber defenders.Status: UnconfirmedVerification trailUnverifiedChecked by verification engine
No atom-level source-text verification is available.
Update log · 0
AI accelerates. Humans approve.