Skip to main content

text model comparison

Gemini 3.1 Flash-Lite vs Mercury 2.5

Gemini 3.1 Flash-Lite and Mercury 2.5 are both current text models, from Google and Inception. Mercury 2.5 is 20% cheaper per 1M input tokens than Gemini 3.1 Flash-Lite, $0.20 vs $0.25. Gemini 3.1 Flash-Lite has the larger context window, 1,048,576 vs 260,000 tokens.

Lower price per 1M input tokens
Mercury 2.5$0.20 against $0.25
Longer context window
Gemini 3.1 Flash-Lite1.05M tokens against 260K tokens
Newer release
Mercury 2.5Sep 2026 against May 2026

Data verified November 3, 2026 · Dataset 2026-11-06

Share this

Side by side

Gemini 3.1 Flash-Lite and Mercury 2.5, line by line

Prices in effect today, compared at the same tier wherever both list one. The better value in each row is highlighted.

FactGemini 3.1 Flash-LiteMercury 2.5
Price
Price per 1M input tokens$0.25$0.20 (better)
Price per 1M output tokens$1.50$0.75 (better)
Independent benchmarks
GPQA Diamond81.8% (#37 of 41)Not measured
FrontierMath Tiers 1-3 (v2)27.7% (#38 of 41)Not measured
Specs
ProviderGoogleInception
Release dateMay 7, 2026September 8, 2026
AccessAPI onlyAPI only
Context window1,048,576 tokens260,000 tokens
Max output65,536 tokens65,536 tokens
Inputtext, image, audio, videotext
Outputtexttext

No independent benchmark has scored both models yet. Each rank counts the current models scored on that benchmark.

Scores:Epoch AI · 2026-11-03

Charts

Against every current text model

Where Gemini 3.1 Flash-Lite and Mercury 2.5 land among the current models of their task, on every independent benchmark, on price and on context.

Where each stands

Only Gemini 3.1 Flash-Lite has a current GPQA Diamond score: #37 of 41. Each bar is the share of current models it beats.

Gemini 3.1 Flash-Lite and Mercury 2.5 among the current text models, one row per benchmark, price and context window.

Independent benchmarks

GPQA Diamond

Gemini 3.1 Flash-Lite:#37of 41

81.8% · ahead of 10%

Mercury 2.5: Not scored

FrontierMath Tiers 1-3 (v2)

Gemini 3.1 Flash-Lite:#38of 41

27.7% · ahead of 8%

Mercury 2.5: Not scored

Price and context

Price per 1M input tokens

Gemini 3.1 Flash-Lite:#10of 58

$0.25 · cheaper than 84%

Mercury 2.5:#7of 58

$0.20 · cheaper than 86%

Price per 1M output tokens

Gemini 3.1 Flash-Lite:#15of 58

$1.50 · cheaper than 75%

Mercury 2.5:#8of 58

$0.75 · cheaper than 88%

Context window

Gemini 3.1 Flash-Lite:#12of 67

1.05M tokens · longer than 67%

Mercury 2.5:#60of 67

260K tokens · longer than 11%

Cost

What each costs

Example workloads at each provider's listed rates, then every other rate either one lists. The cheaper figure in each row is highlighted.

The same work, priced

For 10M input and 2M output tokens, Gemini 3.1 Flash-Lite costs 1.6 times as much as Mercury 2.5, at each provider's listed rates.

What Gemini 3.1 Flash-Lite and Mercury 2.5 cost for the same work, at the rates each provider lists.
  • 10M input and 2M output tokens

    Gemini 3.1 Flash-Lite costs 1.6× as much

    Gemini 3.1 Flash-Lite
    $5.50
    Mercury 2.5
    $3.50
  • The same, with 9M of the input tokens cached

    Gemini 3.1 Flash-Lite costs 1.9× as much

    Gemini 3.1 Flash-Lite
    $3.48
    Mercury 2.5
    $1.88
Workload or rateGemini 3.1 Flash-LiteMercury 2.5
Example workloads
10M input and 2M output tokens$5.50$3.50 (better)
The same, with 9M of the input tokens cached$3.48$1.88 (better)
Other listed rates
1M input tokens, cached$0.025$0.02 (better)
1M input tokens, cached audio$0.05Not listed
1M input tokens, audio$0.50Not listed

Your own bill will differ. These figures use list prices and example workloads. The AI Model Cost Audit reviews your real usage, shortlists models that score as well for each task for less, and estimates the savings.

Launch price $100, regular price $150

Book the cost audit

Verdict

When to choose each

Gemini 3.1 Flash-Lite

Where it leads

  • Larger context window 1,048,576 vs 260,000 tokens
  • Accepts image, audio and video input

Best for

  • High-volume multimodal extraction and translation

Watch out for

  • The published price for audio input differs from the listed text, image, and video input price
  • Google's earliest published shutdown date is 2027-05-07; migrate to gemini-3.5-flash-lite.

Mercury 2.5

Where it leads

  • Lower input and output prices, $0.20 and $0.75 vs $0.25 and $1.50 per 1M tokens
  • Cheaper cached input, $0.02 vs $0.025 per 1M tokens

Best for

  • Latency-critical agent loops and voice agents
  • High-throughput, low-cost text generation

Watch out for

  • Inception publishes no benchmark scores for it, and no independent score is recorded here
  • The recorded rates are list prices; an 80% launch discount was running with no published end date

From benchmark to production

Building on Gemini 3.1 Flash-Lite or Mercury 2.5?

Public scores rarely hold on your own data. We test both models on your real tasks, then build the agent or pipeline around the one that wins.

AI Model Cost AuditLaunch price $100, regular price $150What the audit includes

FAQ

Gemini 3.1 Flash-Lite vs Mercury 2.5 FAQ

Which is cheaper, Gemini 3.1 Flash-Lite or Mercury 2.5?

Mercury 2.5 is 20% cheaper per 1M input tokens than Gemini 3.1 Flash-Lite, $0.20 vs $0.25. For 10M input and 2M output tokens, Gemini 3.1 Flash-Lite costs $5.50 and Mercury 2.5 $3.50 at standard rates.

Which is cheaper with prompt caching, Gemini 3.1 Flash-Lite or Mercury 2.5?

Gemini 3.1 Flash-Lite charges $0.025 per 1M cached input tokens, 90% below its standard input rate, and Mercury 2.5 $0.02, 90% below. With 9M of 10M input tokens cached and 2M output tokens, Gemini 3.1 Flash-Lite costs $3.48 and Mercury 2.5 $1.88.

Which has the larger context window, Gemini 3.1 Flash-Lite or Mercury 2.5?

Gemini 3.1 Flash-Lite has the larger context window, 1,048,576 vs 260,000 tokens. Both return up to 65,536 tokens per response.

How do Gemini 3.1 Flash-Lite and Mercury 2.5 differ in features?

Gemini 3.1 Flash-Lite takes text, image, audio and video input; Mercury 2.5 takes text.

Production evaluation

Choosing between Gemini 3.1 Flash-Lite and Mercury 2.5?

Public benchmarks narrow the shortlist. We run both on your own tasks, data and cost limits, then build the routing and review controls to run the winner in production.

  • A head-to-head evaluation on your real workload
  • Routing between models by cost and difficulty