Back to programming-model index

MODEL EVIDENCE / OpenRouter #17

Google: Gemini 3 Flash Preview

The model identity, provider record, and independently verified benchmark results in one place. The number above is OpenRouter discovery order, not benchmark rank.

google/gemini-3-flash-preview
Official source pendingComplete result
Fast numbersCaptured facts and latest status
OpenRouter feed rank
OpenRouter #17
Context
1048576
Input
text, image, file, audio, video
Output
text
Reasoning flag
Yes
Prompt price
$0.500 / 1M tokens
Completion price
$3.000 / 1M tokens
Latest run
2026-08-31T21:35:59.284247+00:00
View denominator
50

DEEP BENCHMARK

What the run established.

Counts stay visible for auditability. A score appears only when the Most Annoying 50-trap view satisfies the independent checks.

MOST ANNOYING BENCHMARK

22/ 50

Complete result · 22 verified mistakes · 28 clean results · 0 unresolved

Inspect run receipt
Packs & trend

Pack 1: 3 verified mistakes, 7 clean results, 0 unresolved; Pack 2: 4 verified mistakes, 6 clean results, 0 unresolved; Pack 3: 4 verified mistakes, 6 clean results, 0 unresolved; Pack 4: 6 verified mistakes, 4 clean results, 0 unresolved; Pack 5: 5 verified mistakes, 5 clean results, 0 unresolved

2026-08-28T05:52:59.444809+00:00: 13; 2026-08-31T21:35:59.284247+00:00: 22

Score history

Model runs, then future slots.

  1. 0174%
  2. 0256%
  3. 03NOT RUN
  4. 04NOT RUN
  5. 05NOT RUN
  6. 06NOT RUN
  7. 07NOT RUN
  8. 08NOT RUN
  9. 09NOT RUN
  10. 10NOT RUN
  11. 11NOT RUN
  12. 12NOT RUN
  13. 13NOT RUN
  14. 14NOT RUN
  15. 15NOT RUN
  16. 16NOT RUN
  17. 17NOT RUN
  18. 18NOT RUN
  19. 19NOT RUN
  20. 20NOT RUN
  21. 21NOT RUN
  22. 22NOT RUN
  23. 23NOT RUN
  24. 24NOT RUN
Verified mistakesClean resultsUnresolved / pendingMissing / no recorded result

Most Annoying verified mistake receipts

Only receipts from the Most Annoying view are listed here. Tolerable-view receipts are separate and are not added to this score.

  1. Precedence TraceT002

    The model's exact answer did not match the frozen expected output.

    sha256:6200351d4ca4a6e69c78c1d7538197a6312c4420393e75be38fbb16b3a9ef44f
  2. Slice TraceT004

    The model's exact answer did not match the frozen expected output.

    sha256:503adb7811c3db344bc559738b9fee759c93b48bd1586e1f95484581ea517b03
  3. Alias TraceT005

    The model's exact answer did not match the frozen expected output.

    sha256:ad4f1570bc21f0bd688f2fbba791c21854c6383212621bd9015deb99af66f106
  4. Precedence TraceT022

    The model's exact answer did not match the frozen expected output.

    sha256:aef6d8800c2c9a8288821d3390a355a7f76039041a1ce1cbf0099da5da4df8c9
  5. Slice TraceT024

    The model's exact answer did not match the frozen expected output.

    sha256:63348a4a383133408bbc986109b445642131a3ddcff41d6e8f09b3eb79155e45
  6. Precedence TraceT026

    The model's exact answer did not match the frozen expected output.

    sha256:83ec56c476b18961c0eada110511e7185b4a50fc32115b8bc9b95726fd1a14b5
  7. Precedence TraceT030

    The model's exact answer did not match the frozen expected output.

    sha256:7e5bf373714adc2e09a365695b7d6b03126c115fc1bc91b47a225a2acb092ee8
  8. Alias TraceT041

    The model's exact answer did not match the frozen expected output.

    sha256:bd44d33c961313aefabe05c4a6293a150522ff6e603eaaf86ece0b981ac3fca4
  9. Precedence TraceT042

    The model's exact answer did not match the frozen expected output.

    sha256:2c9cbb7ca8cf3f6ea710320cc2e7d071b9e240b35e04f5d6f0b06878ca0de18d
  10. Alias TraceT049

    The model's exact answer did not match the frozen expected output.

    sha256:42925e81a1b8dc4e713faf0c99d5ff5fd3b684d4747ebfc8e1be150c99a6b253
  11. Precedence TraceT050

    The model's exact answer did not match the frozen expected output.

    sha256:bc7ecef98d6568031d42923089d4149d3e0a9c43de6540bd393c42744452a922
  12. Precedence TraceT062

    The model's exact answer did not match the frozen expected output.

    sha256:c0c846d094e648d12dd5eaf0743de89b790b7a0ca09a0b10f67cc5c72a53480f
  13. Slice TraceT064

    The model's exact answer did not match the frozen expected output.

    sha256:ef58d389f2eb43cfd97c0e7190574fbdc67e67bdebcc32517aa1a32ca5eb7a1e
  14. Alias TraceT065

    The model's exact answer did not match the frozen expected output.

    sha256:239dd4e2d90632aaa7104a2dab95f03f7ab1ff5332b700788a36d8644c0f928e
  15. Precedence TraceT066

    The model's exact answer did not match the frozen expected output.

    sha256:02377fce967fa7c55025be5a4e8fd6d7f29582a6d11bbc3b31d83cd01d16ee15
  16. Slice TraceT068

    The model's exact answer did not match the frozen expected output.

    sha256:d3796419ac6bb03f8920f4d10d66353b2014c30e9528aefddfa2259b12e96e7e
  17. Precedence TraceT070

    The model's exact answer did not match the frozen expected output.

    sha256:873c608cd7d83da9d4d925fae07a5aad92bdc458165152bca4be98d6dacdf0c8
  18. Slice TraceT084

    The model's exact answer did not match the frozen expected output.

    sha256:fd1c30c7d6c042d5f23a8d517a6b484eb69ddf2decd8dbac3a29075ec3135e13
  19. Alias TraceT085

    The model's exact answer did not match the frozen expected output.

    sha256:bc91d66822527d1662bdddeda914a06d4cdc95cec2a80c560167991b43b5b1ba
  20. Precedence TraceT086

    The model's exact answer did not match the frozen expected output.

    sha256:030b8a0949dd1f9c906fd6692eaa464f528c298367f308a041c74335caa2e975
  21. Alias TraceT089

    The model's exact answer did not match the frozen expected output.

    sha256:8bb9e574f4b6452b13590efd22222273c5849fb9972648b3c7ff6cf2d5d9ee0b
  22. Precedence TraceT090

    The model's exact answer did not match the frozen expected output.

    sha256:1d518257b3805cbb3930ede46217df380d29f1cc63085719a9c2d5ddcb291082

OFFICIAL SOURCES

Provider record.

An exact primary official launch or model-card source has not yet been verified for this captured model ID.

STATUSOfficial source pendingVerified: PendingResearch method: official primary-web fallback
  • No exact primary launch or model card is certified yet. This page will not infer one from a similar name.

Important: Statements in linked publisher material are provider claims. They are not findings of this benchmark.

METHOD & LIMITS

What this page can say.

Every public leaderboard score is the latest Most Annoying 50-trap view. The 100-task run receipt is the wider audit bundle, not the benchmark denominator. Benchmark rank #1 means fewest verified mistakes among complete runs; the list can still be sorted by most mistakes for readability. OpenRouter feed rank is separate discovery metadata and is kept out of the benchmark rank column. Two-week average reporting will use the last four complete 100-task cycles when enough cycles exist. Individual verified mistakes, called catches internally, require primary and shadow checks to reproduce the exact mismatch from the recorded reply against the frozen expected output. Retrieval is not signature verification; auditors should recompute receipt roots and verify signatures against the published issuer key. Routing errors, timeouts, incomplete answers, and missing evidence stay outside the ranked score. Do not judge any model solely from this benchmark. Ranked complete runs must use the same 50 traps, scoring rules, model settings, and retry policy.

  1. Dual-check reproduction

    For each listed Most Annoying mistake, primary and shadow verification must both reproduce the exact mismatch from the recorded reply against the frozen expected output. Otherwise the item remains unresolved.

  2. Certified denominator

    The rankable score is the 50-trap Most Annoying view. The run receipt may say 100 tasks because it covers the wider audit bundle; that object is evidence, not a separate leaderboard denominator.

  3. Signature and feed rank

    Independent verification means recomputing the receipt root and checking the signature against the published issuer public key. OpenRouter rank stays discovery metadata and does not affect benchmark rank.

Reproduce this result

  1. Inspect the run receipt.
  2. Inspect each Most Annoying mistake receipt.
  3. Verify each receipt root and signature.
  4. Wait for the full proof bundle if it is not released yet.
  5. Rerun the exact prompt and expected-output comparison.

Claim ceiling: OpenRouter rank and metadata describe the captured discovery feed. Provider claims are separate from this benchmark. Scores describe only the most recent public test pack and should not be used alone to judge the model generally.

Feed root
sha256:61e4106b1d08b7ef16ec9d332548214d01937a628e573fcf9a3091f601dfcb76
Model metadata root
sha256:954aabe05e1a463d7c193ab83ec3281906f58d54ca4504efd1f7ecb8cab9dab6
Official-source registry root
sha256:e5606a5133065340a402faa6c24a6bb3fb36d82809003a75bae28e85d7ec8e68

QUOTE THE RECORD

Share the exact page, not a screenshot without context.

Discuss this result in:

GET A SECOND EXPLANATION

Ask your favorite AI what this means.

Copy a bounded prompt that includes this model page and its claim ceiling.