Skip to main content
Meo

Google: Gemini 2.5 Flash Lite

Google: Gemini 2.5 Flash Lite is google's model. On the un-leaked meo benchmark it ranks #50 of 82 with 57.3% accuracy and an Effective Value (𝕍) of 0.0054 β€” earning a correct answer for about $0.0037 in 44s each.

as of 2026-10-18

meo rank
#50
of 82
Accuracy
57.3%
overall
𝕍 index (top=100)
0.13
raw 𝕍 0.0054 Β· N=10
$ / correct
$0.0037
44s each

Per-domain accuracy

145 correct of 250 questions asked across 11 domains Β· answer rate 100% Β· macro accuracy 56.6%. Headline accuracy is micro-averaged per meo’s methodology; per-domain question counts (correct/asked) are shown so you can judge significance.

Theory of mind
100%3/3
Tape machine
87%13/15
Regex automaton
81%13/16
Watson-Glaser
75%3/4
Framework bias
67%2/3
Base-rate bias
63%26/41
Logic / math / CS
59%20/34
Perceptual illusions
56%57/102
CSP (unsatisfiable)
50%8/16
State tracking
0%0/1
Long arithmetic
0%0/15

Efficiency

Cost per correct
$0.0037
Seconds per correct
44
Tokens per correct
9,056
Correct answers
145 / 250
Total benchmark cost
$0.54
Avg tokens / item
5,253

Model details

Released
2025-07-22
Weights
β€”
Context
1.0M
Max output
66K
Knowledge cutoff
2025-01-31
Price in
$0.10/M
Price out
$0.40/M
Modality
text+image+file+audio+video->text
View on OpenRouter β†—

Sources: openrouter Β· meo

FAQ

How accurate is Google: Gemini 2.5 Flash Lite?

Google: Gemini 2.5 Flash Lite scores 57.3% overall on the meo benchmark (rank #50 of 82), an un-leaked private holdout graded objective-first.

What is Google: Gemini 2.5 Flash Lite's Effective Value?

Its Effective Value (𝕍) is 0.0054 at N=10, Ο‰=1, Ξ΄=1.5 β€” a fused metric of accuracy, speed, cost, and error-cascade. 𝕍 rankings shift with task depth.

How much does Google: Gemini 2.5 Flash Lite cost per correct answer?

About $0.0037 per correct answer (9,056 tokens, 44 seconds each) at benchmark-time pricing via OpenRouter.

Is Google: Gemini 2.5 Flash Lite open weights?

Open-weights status for Google: Gemini 2.5 Flash Lite is not recorded.

Benchmark updates affecting this model

  • 2026-06-09majorlogic_math_cs ground truth was broken β€” culled and rebuilt with solver-verified items

    A 5-lab frontier jury revealed that 29 of 37 LLM-authored logic_math_cs items had broken ground truth: on most, five independent frontier models unanimously concluded the clue sets were contradictory while the stored answer claimed a number, and on eight, all five agreed on the SAME different number (author arithmetic errors). Models were being marked wrong for being right. The 29 were retired (full audit published in the repo) and the domain rebuilt with generator-as-oracle items: an embedded constraint solver PROVES each puzzle has exactly one solution (or provably none), verified by an independent exhaustive enumerator. A 3-round difficulty study found frontier models saturate consistent human-scale logic puzzles, so items use anonymous global constraints and carry a measured difficulty gradient.

  • 2026-06-08majorCorrected base_bias grading (LLM-equivalence judge fix)

    The LLM-equivalence grader uses a reasoning model whose verdict token budget was too small, so its YES/NO answer was emitted as empty and silently counted as a failure. This understated the base_bias domain to ~7% for every model when the true accuracy is ~85-90%. We raised the judge budget, excluded its reasoning tokens from the verdict, switched to verbosity-robust conclusion matching, and re-graded the whole board.

Full benchmark changelog β†’

Embed this ranking

Show Google: Gemini 2.5 Flash Lite’s meo rank on your site β€” the badge links back here.

Related: all Google models Β· Theory of mind domain Β· best Effective Value Β· biggest movers Β· methodology Β· full leaderboard.