The newest Gemini Pro model is Gemini 4 Argon, released on October 13, 2026. It shipped 11 days ago, and the line's usual gap is about 236 days.
How long it has been since the last release against the line's usual gap and its recent ones — the rhythm, not a date.
Latest: Gemini 4 Argon · Oct 13, 2026
Every Monday, only when something shipped: the AI models that came out, what each one changed, and which lines are due next.
We email you a link to confirm first · unsubscribe anytime · read past issues · about the newsletter · RSS for the Gemini Pro line
Newest first — what each version changed versus the one before, in Google's own numbers.
Google's first Gemini 4 model replaces Gemini 3.1 Pro as its top model in the paid Gemini API, with a 1M-token window shared by input, thinking and output.
Compared with Gemini 3.1 Pro · the model it replaces
Answers nearly 9 in 10 questions about reasoning in the physical world
ERQA — embodied reasoning questions for robots and real scenes
87%
✦ Google gives no earlier Gemini score to compare
Scores 53 on a combined set of hard tests
Artificial Analysis Intelligence Index v4.3.2 — one score averaging 10 hard tests
53
✦ High effort · no Gemini 3.1 Pro score on the same index version to compare
Launches at an introductory price for developers
API price — dollars per 1M input / output tokens
$2 / $10
✦ introductory price · Google's standard rate is $4 / $20
Google's updated flagship: its score on brand-new reasoning puzzles more than doubled over Gemini 3 Pro — at the same price.
Compared with Gemini 3 Pro · the model it replaces
Solves about 8 in 10 brand-new reasoning puzzles
ARC-AGI-2 — novel puzzles a model hasn't seen before
77.1%
▲ up from 31.1% · more than doubled
Handles 7 in 10 tasks that use outside tools
MCP Atlas — calling tools and services to get work done
69.2%
▲ up from 54.1% · +15 points
Costs the same to run as before
Price per 1M tokens — what developers pay, input / output
$2 / $12
same price as Gemini 3 Pro
Gemini 3 Pro fixes more real software bugs than Gemini 2.5 Pro (76.2% vs 59.6% on SWE-bench Verified) but costs more to run: $2 / $12 per 1M tokens, up from $1.25 / $10.
Compared with Gemini 2.5 Pro · the model it replaces
Fixes about 3 in 4 real software bugs
SWE-bench Verified — fixing real bugs in real software projects
76.2%
▲ up from 59.6% in Google's Gemini 3 table · +16.6 points
Costs developers more to run than Gemini 2.5 Pro
Price per 1M tokens — what developers pay, input / output (prompts up to 200K)
$2 / $12
up from $1.25 / $10 · preview price at launch
Gemini 2.5 Pro is a thinking model that reasons before it answers and scored 18.8% on Humanity's Last Exam without tools, but it launched with a 1M-token context window, half of Gemini 2.0 Pro's 2M.
Compared with Gemini 2.0 Pro · the model it replaces
Answers almost 1 in 5 expert-level exam questions without outside help
Humanity's Last Exam — very hard expert questions, answered without tools
18.8%
✦ best score without tools at launch, Mar 2025 (Google)
Google's most expensive model for developers when paid use opened
Price per 1M tokens — what developers pay, input / output (prompts up to 200K)
$1.25 / $10
✦ paid price set Apr 2025 · above every other Google model at the time
Gemini 1.5 Pro beat Gemini 1.0 Pro on 87% of the benchmarks Google uses to develop its models and found a planted fact 99% of the time in inputs up to 1 million tokens long.
Compared with Gemini 1.0 Pro · the model it replaces
Reads up to 1 million tokens at once — over 700,000 words or 1 hour of video
Context window — how much text, audio or video the model can take in at once
1 million tokens
▲ up from 32,000 tokens for Gemini 1.0