Applied artificial intelligence, neural systems, and smart agents.
Fusing generative AI, multi-agent frameworks, and high-throughput data processing to automate business workflows.
Gemini 3.1 Pro
Gemini 3.1 Pro ships under the identifier gemini-3.1-pro-preview and its pricing structure is tiered: below the 200,000 token threshold, $2 per million input tokens and $12 per million output; at 200,000 tokens and above, $4 and $18. It carries no rank in our coding table because eighty-five of a hundred weighted points lack verified evidence.
- preview-status identifier
- pricing threshold at 200k tokens
- unranked in our table
Last checked: This is a reference page. It is re-checked against the vendor sources and updated when a new version ships. Change log
Tiered pricing structure, with the threshold at 200,000 tokens
Most models in this category run a single flat rate. Gemini 3.1 Pro defines two pricing tiers. Below the 200,000 token threshold, the input rate is $2 and the output rate is $12 per million tokens. At 200,000 tokens and above, these figures shift to $4 and $18: input doubles, output rises by half again.
One point in Google's pricing specification remains undocumented, and we do not estimate it: whether crossing the threshold re-rates the entire request or only the portion above it. The specification lists only the two tiers and does not describe the transition mechanism. For architectures that construct long requests, this ambiguity has a measurable effect on billing and should be verified against a live call.
Why it carries no rank in the coding table
The ranking rule is explicit: a candidate missing verified evidence for more than half the criteria weight falls outside the ranked table and is listed independently. Gemini 3.1 Pro is missing eighty-five of one hundred weighted points, the only candidate in this position.
Three evidence gaps are traceable, and each is fixable. This model's row on the WebDev Arena leaderboard is not recorded, and that column alone carries forty points of weight. Google's models page publishes no context window figure for this version, so the ten-point context window column has no value; we do not record a figure absent from the primary source. A tooling maturity score has not been assigned to this version either, accounting for fifteen more points. The only evidence on record today is a single row on the general text leaderboard at a score of 1486.
This evidence gap should not be read as a deficiency in the model itself. A table that fills an empty cell with a plausible figure is less reliable than one that reports the gap explicitly.
Preview status, and Iran's access coverage
The identifier Google has published terminates in preview, meaning the deployment target has not been declared final by its vendor. The identifier can change, model behaviour can shift without notice, and pricing can differ in the stable release. This is an acceptable risk profile for prototyping; it is not acceptable for infrastructure expected to remain operational six months out.
This page's Iran access column is derived from the Gemini API available-regions list. That list names more than 190 countries and territories, and Iran is not among them. A pattern is visible across the five models documented this week: only Google and Anthropic publish a regions list of this kind. Moonshot, Z.ai and DeepSeek publish none, so the Iran column for those three remains in an unverified state. A vendor that documents its operating restrictions is more transparent than one that documents nothing.
Using it from Iran
This column carries a date and says how it was checked, because most listicles guess it. Where we have only read a vendor policy page, the note below says exactly that.
- Reachable
- blocked
- Payment
- no working route
- Free tier
- no
- Measured on
How we checked: The Gemini API available-regions specification defines the set of countries where the API is operational, and Iran is not on that list. This status was extracted from the specification itself rather than a network measurement; an independent test has not yet been run.
Documented strengths
- The first pricing tier, $2 and $12, undercuts both Opus 5 and Fable 5.
- It records a score of 1486 on the general text leaderboard, within the top cluster of the table.
- Google publishes an available-regions list for the API, so the access column is not derived from assumption.
Room for improvement and technical considerations
- Carries no rank in the coding table: eighty-five of a hundred weighted points lack verified evidence.
- Google publishes no context window figure for this version on its models page, so none is recorded here.
- Its row on the web-dev leaderboard has not been read, and that column accounts for forty points of the table's weight.
- Its identifier terminates in preview, so behaviour and pricing can change without notice.
- The rate-application mechanism above the 200k token threshold — full request or marginal tokens only — is not documented.
- Iran is not on the Gemini API available-regions list.
Technical verdict
While the identifier remains in preview status and the context window specification is unpublished, we do not recommend deploying this model as the foundation of a production system. It is suitable for testing and comparison. If Google publishes a web-dev leaderboard row and a context window figure, it re-enters the ranked table the same day and earns its own rank.
Frequently raised questions, with documented answers
What does Gemini 3.1 Pro cost
Below the 200,000 token threshold, $2 per million input tokens and $12 per million output. At 200,000 tokens and above, $4 and $18.
What is the Gemini 3.1 Pro context window
Google does not publish that figure on the Gemini API models page, so none is recorded in this reference. Any number seen elsewhere was not drawn from the primary source.
Why is it unranked in your coding table
Because eighty-five of a hundred weighted points lack verified evidence: the web-dev leaderboard row, the context window, and a tooling score. Our rule leaves a candidate unranked when more than half the criteria weight is empty.
What does a preview identifier mean
That Google has not yet declared this version stable. The identifier, its behaviour, and its pricing can all change before the final release, a risk profile for any system with a multi-month operating life.