Skip to main content
Meo

ByteDance Seed: Seed 1.6 Flash

ByteDance Seed: Seed 1.6 Flash is bytedance-seed's model. On the un-leaked meo benchmark it ranks #47 of 82 with 57.7% accuracy and an Effective Value (𝕍) of 0.0034 β€” earning a correct answer for about $0.0022 in 53s each.

as of 2026-10-18

meo rank
#47
of 82
Accuracy
57.7%
overall
𝕍 index (top=100)
0.080
raw 𝕍 0.0034 Β· N=10
$ / correct
$0.0022
53s each

Per-domain accuracy

144 correct of 250 questions asked across 11 domains Β· answer rate 100% Β· macro accuracy 59.0%. Headline accuracy is micro-averaged per meo’s methodology; per-domain question counts (correct/asked) are shown so you can judge significance.

Regex automaton
81%13/16
Logic / math / CS
76%26/34
CSP (unsatisfiable)
69%11/16
Tape machine
67%10/15
Framework bias
67%2/3
Theory of mind
67%2/3
Base-rate bias
63%26/41
Perceptual illusions
50%51/102
Watson-Glaser
50%2/4
Long arithmetic
7%1/15
State tracking
0%0/1

Efficiency

Cost per correct
$0.0022
Seconds per correct
53
Tokens per correct
7,315
Correct answers
144 / 249
Total benchmark cost
$0.32
Avg tokens / item
4,230

Model details

Released
2025-12-23
Weights
β€”
Context
262K
Max output
33K
Knowledge cutoff
β€”
Price in
$0.07/M
Price out
$0.30/M
Modality
text+image+video->text
View on OpenRouter β†—

Sources: openrouter Β· meo

FAQ

How accurate is ByteDance Seed: Seed 1.6 Flash?

ByteDance Seed: Seed 1.6 Flash scores 57.7% overall on the meo benchmark (rank #47 of 82), an un-leaked private holdout graded objective-first.

What is ByteDance Seed: Seed 1.6 Flash's Effective Value?

Its Effective Value (𝕍) is 0.0034 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 ByteDance Seed: Seed 1.6 Flash cost per correct answer?

About $0.0022 per correct answer (7,315 tokens, 53 seconds each) at benchmark-time pricing via OpenRouter.

Is ByteDance Seed: Seed 1.6 Flash open weights?

Open-weights status for ByteDance Seed: Seed 1.6 Flash 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 ByteDance Seed: Seed 1.6 Flash’s meo rank on your site β€” the badge links back here.

Related: all Bytedance Seed models Β· Regex & automata domain Β· best Effective Value Β· biggest movers Β· methodology Β· full leaderboard.