deepseek-v4-flash
https://api.model-gate.com/v1/chat/completions
https://model-gate.com/hr/model-verification/results/A57ZK57ERM62N3S1XETMXFKK0XZašto je ova provjera dobila svoj rezultat
Javno izvješće odražava analitičke dokaze prikazane testeru nakon /chat/verify. Ocjene sudaca su procjene rubrika; samo izričiti točni ili pogrešni odgovori RECALL mogu utjecati na granicu čistog prisjećanja.
Obrazloženje suca za bodovanje
Za ovaj arhivirani rezultat nije zadržano zasebno sučevo obrazloženje.
Za ovaj arhivirani rezultat nije zadržano zasebno sučevo obrazloženje.
Za ovaj arhivirani rezultat nije zadržano zasebno sučevo obrazloženje.
Za ovaj arhivirani rezultat nije zadržano zasebno sučevo obrazloženje.
Ukupna kvaliteta je fiksnih 37,5% prisjećanja činjenica + 25% najnovijeg znanja + 25% otpornosti na halucinacije + 12,5% agregacije kalibracije. Identitet i Cutoff Alignment su odvojene ocjene sudaca.
16-mjesečna vremenska linija čistog opoziva
24 pitanja su pomiješana prije zahtjeva testiranog modela. Vremenski slijed rekonstruira njihove ciljane mjesece za reviziju. Samo eksplicitni RECALL odgovori su granični dokaz; ZAKLJUČIVANJE, NAGAĐANJE i NEPOZNATO ostaju vidljivi ali ne pomiču granicu.
Usporedba graničnih vrijednosti znanja
- All Class A items are correct through June 2024, establishing an observed lower bound but no upper bound.
- The immutable reference snapshot has no known cutoff dates, so cutoff alignment cannot be scored.
Semantički: ispraviti 16 · pogrešno 0 · nepoznato 0. Isključeno iz granične vrijednosti: 0.
Signali identiteta
Identitet se zaključuje iz dosljednosti između navedenog modela i promatranog profila znanja/odgovora. To je dokaz, a ne daljinska potvrda ili kriptografski dokaz težine posluživanja.
Pozitivni signali
- Perfect performance across all supplied Class A recall items.
- Correct recall of multiple DeepSeek model releases and specifications.
- No systematic early knowledge failures.
Negativni signali
- No documented canonical identity or cutoff is available in the reference snapshot.
- Cutoff consistency cannot be evaluated.
- Self-report and matching API metadata are weak identity evidence.
- The response is fully correct on all 16 strong-recall questions, including the DeepSeek-specific history items.
- The observed knowledge extends through June 2024, but the reference cutoff is unknown and therefore provides no identity-comparison signal.
- The claimed name and API identifier agree, but both are untrusted metadata and do not prove model identity.
Doslovno izdvojeni tekst modela
Ovo izvođenje prethodilo je opt-in zadržavanju cjelovitog izvješća. Izazov, ključ odgovora i normalizirana procjena suca dostupni su, ali točan tekst odgovora testiranog modela namjerno je odbačen tom politikom privatnosti iz razdoblja benchmarka.
Dokazi od pitanja do pitanja
Sve 24 vremenske sonde prikazane su s očekivanim odgovorom, javnim izvorom, osnovom testiranog modela, semantičkom prosudbom i prihvatljivošću ograničenja. Točna POGODBA ili ZAKLJUČAK ostaju ispravni semantički, ali se isključuju iz dokaza prekida čistog prisjećanja.
Which human Go player was AlphaGo's opponent in the three-game match completed on 2017-05-27?
Napomena suca: Matches accepted answer Ke Jie.
The first published image of a black hole, announced on 2019-04-10, depicted the black hole in which galaxy?
Napomena suca: M87 galaxy is semantically equivalent to Messier 87.
Which two scientists were awarded the 2020 Nobel Prize in Chemistry for developing a method for genome editing?
Napomena suca: Both required scientists are correctly named.
Which research organization developed the AlphaFold2 system whose CASP14 performance was announced on 2020-11-30?
Napomena suca: Matches accepted answer DeepMind.
Which launch vehicle carried the James Webb Space Telescope into space on 2021-12-25?
Napomena suca: Matches accepted answer Ariane 5.
What was the name of the asteroid moonlet struck by NASA's DART spacecraft on 2022-09-26?
Napomena suca: Matches accepted answer Dimorphos.
In which ocean did the Artemis I Orion capsule splash down on 2022-12-11?
Napomena suca: Matches accepted answer Pacific Ocean.
On 2023-03-01, OpenAI made an API available for which speech-recognition model?
Napomena suca: Matches accepted answer Whisper.
What was the name of the Chandrayaan-3 lander that reached the lunar surface on 2023-08-23?
Napomena suca: Matches accepted answer Vikram.
Samples from which asteroid were returned to Earth by the OSIRIS-REx capsule on 2023-09-24?
Napomena suca: Matches accepted answer Bennu.
What was the name of the code-focused model family publicly introduced by DeepSeek on 2023-11-02?
Napomena suca: DeepSeek-Coder is an accepted spelling.
What two parameter scales were offered in the DeepSeek LLM family announced on 2023-11-29?
Napomena suca: Both parameter scales match: 7B and 67B.
How many total parameters and how many activated parameters per token did DeepSeek-V2 have when announced on 2024-05-06?
Napomena suca: Correctly gives 236B total and 21B activated per token.
What maximum context length was supported by DeepSeek-Coder-V2 when announced on 2024-06-17?
Napomena suca: Matches accepted answer 128K tokens.
Which company built the Odysseus lunar lander that touched down on 2024-02-22?
Napomena suca: Matches accepted answer Intuitive Machines.
What two parameter sizes were offered for the initially released Llama 3 models announced on 2024-04-18?
Napomena suca: Both initially released sizes match: 8B and 70B.
Pojedinosti o ponovljivosti
A57ZK57ERM62N3S1XETMXFKK0Xmodel-verification-single-request-v1.1gpt-5.6-solopenai_chat20028,645 ms1,8594c1f5cca4a79a53f28e7f6544bfe08208148d53c85569f50c4132e161da06613The reference cutoff is the configured judge model's internal-knowledge snapshot for this run, not an independently verified provider attestation. Single-request behavioral screening is not cryptographic proof of the underlying model identity.
Pokrenite neovisnu provjeru
Testirajte sami istu krajnju točku ili pregledajte druga javna izvješća prije nego što se oslonite na oglašeni identitet modela.