GPT-Rosalind vs Qwen 4 Max: 3.000 Agen, Satu Tagihan, dan Model yang Belum Dirilis

Thirty-two minutes, and more than 700 references. On 21 October 2026 O​penAI published a collaboration with PerturbAI and Scripps Research in which thousands of agents running on GPT-Rosalind worked through a whole-brain CRISPR atlas of millions of mouse cells, each agent taking one gene disruption, searching the literature around it, and writing a report — a job the authors estimate at months to a year done one perturbation at a time. The model itself is not new: O​penAI introduced GPT-Rosalind on 16 April 2026, widened it beyond a named partner list on 11 September, and began billing at $5.00 per million input tokens and $25.00 per million output on 5 October. What is new is the first public look at the model running as a fleet rather than as a call, and that changes what a comparison with Q​wen 4 Max should be about. Q​wen 4 Max was announced at A​libaba's Apsara conference on 22 September 2026 and still has no price, no context window and no model card. So the honest version of this matchup has three sides: a specialist whose real unit cost is per agent, an opponent that has not shipped, and the Q​wen model you can actually rent today — Qwen3.8-Max, at $2.00/$6.00 per million tokens. Announcements that never become a callable endpoint are recorded with their dates and sources in the AI Incident Archive, alongside the rest of the model and vendor incidents we follow.

Apa yang sebenarnya ditunjukkan oleh proses 21 Oktober

Kumpulan data ini adalah atlas Perturb-seq: gen-gen diganggu satu guide RNA pada satu waktu di seluruh otak tikus, dengan transkriptom setiap sel dibaca setelahnya. Keluarannya adalah matriks berisi ribuan perturbasi terhadap jutaan sel, jauh lebih banyak daripada yang dapat dibaca dengan cermat oleh laboratorium mana pun — inilah hambatan yang coba diatasi oleh makalah tersebut.

The method, which the authors call distributed biological reasoning, is not one large model call over a big table. Each agent receives a single gene disruption, examines the cellular response, retrieves the surrounding literature, and produces a structured report proposing a mechanism. Those reports are then assembled into what the paper calls a reasoning atlas: a graph linking genes, cell-type effects, papers, and candidate mechanisms. The runtime O​penAI reports for the whole sweep is 32 minutes.

Temuan yang paling penting bagi siapa pun yang memilih model bukanlah biologi. Yang penting adalah kontrol yang dijalankan tim terhadap metode mereka sendiri, karena yang satu itu memiliki angka yang menyertainya. Mereka menjaga model dan komputasi tetap sama dan memberikan seluruh kumpulan data kepada satu agen alih-alih membaginya ke ribuan agen. Seiring bertambahnya beban kerja, cakupan literatur agen tunggal itu runtuh: pada beban kerja terbesar, agen itu mengutip 11 referensi dibandingkan dengan lebih dari 700 untuk agen terdistribusi. Pakar manusia yang dibutakan lebih menyukai laporan terdistribusi 75% dari waktu.

Read those together and the result is about decomposition, not about model IQ. A single context window degrades when you pour an atlas into it; a few thousand narrow agents with a retrieval tool each do not. The authors' own framing supports this — they present the DCTN1 cholesterol-transport hypothesis as a question for follow-up experiments, not a discovery — and the paper is a preprint posted on 16 October 2026, co-authored by O​penAI and PerturbAI employees, with no peer review yet. Treat the 32 minutes as a real runtime and the human comparison as an order of magnitude.

Untuk halaman versus, konsekuensinya bersifat spesifik. Jika kemenangan itu berasal dari arsitekturnya, maka sifat penentu dari model yang mendasarinya bukanlah skor benchmark-nya — melainkan berapa biaya untuk memanggilnya tiga ribu kali dalam setengah jam, dan apakah Anda diizinkan untuk melakukannya.

Angka yang menentukan pertarungan ini adalah biaya per agen

Published comparisons of these two models tend to line up headline rates and stop. Headline rates are the wrong unit for the workload the 21 October paper just made concrete, because a distributed run multiplies the model call by the number of perturbations. Work it through with stated assumptions: an agent that reads a shared atlas preamble, pulls a handful of papers, and writes a report might consume 60,000 input tokens and 4,000 output tokens. At O​penAI's published rates that agent costs $0.40 — $0.30 of input and $0.10 of output. The same agent on Qwen3.8-Max at $2.00/$6.00 costs $0.14. Three thousand agents: about $1,200 against about $432.

Cache the shared prefix and the gap widens rather than closes, because the cached-input rates diverge more than the headline ones. Put 50,000 of those 60,000 input tokens in a cached prefix and the Rosalind agent drops to roughly $0.18 while the Q​wen agent drops to roughly $0.06 — a little over three times cheaper, because cached input is $0.50 per million on one side and $0.25 on the other. That is our arithmetic, not a vendor figure, and it is sensitive to how much of your prompt is genuinely shared; the direction is not.

Adu keduanya satu sama lain pada dimensi yang menentukan jalannya operasi armada:

• Input price — GPT-Rosalind $5.00 per 1M vs Qwen3.8-Max $2.00. Q​wen 4 Max: not published.

• Output price — GPT-Rosalind $25.00 per 1M vs Qwen3.8-Max $6.00. Q​wen 4 Max: not published.

• Cached input — GPT-Rosalind $0.50 per 1M, and O​penAI's pricing page states cache-write pricing does not apply to the model; Qwen3.8-Max $0.25. Q​wen 4 Max: not published.

• Cost per 3,000-agent sweep — about $1,200 on GPT-Rosalind vs about $432 on Qwen3.8-Max at the assumptions above. Q​wen 4 Max: not computable.

• Context — 1M tokens on both sides. Q​wen 4 Max: not published, and specifically not inherited from the Qwen3.8 window by virtue of the tier name continuing.

• Output maksimum — GPT-Rosalind tidak mempublikasikan batas per respons; snapshot Qwen3.8-Max 0902 mendokumentasikan 131.072 token. • Latensi — Qwen3.8-Max mencatat waktu ke token pertama p50 sebesar 6,32 detik di OrcaRouter dan 40,3 token output per detik pada indeks Artificial Analysis. GPT-Rosalind tidak mempublikasikan angka latensi, yang untuk proses fan-out justru merupakan angka yang paling Anda butuhkan.

• Skor independen — GPT-Rosalind sama sekali tidak dilacak pada Artificial Analysis Intelligence Index karena merupakan model domain dengan akses terbatas; Qwen3.8-Max mencatat 45, berada di peringkat ke-33 dari 214 model, dengan biaya $5.41 per tugas indeks yang diselesaikan dan 40.3 token keluaran per detik.

• Access — GPT-Rosalind requires clearance through O​penAI's trusted-access programme; Qwen3.8-Max is a normal commercial API with an open sign-up.

Two of those lines do more work than the rest. The independent score is the only number in the table nobody involved in either model supplied, and it says the Q​wen flagship is a competent mid-frontier generalist rather than a frontier leader — 45 against 58 for the best model on that board. And the access line is not a footnote: it decides whether the fleet can exist at all.

Where Q​wen 4 Max actually stands

A​libaba used its Apsara conference on 22 September 2026 to name four tiers — Q​wen 4 Max as the flagship, Q​wen 4 Flash for throughput, Q​wen 4 Plus in the middle, and an open-weight Q​wen 4 27B — described the family as training on a new-generation architecture, and said it would arrive soon. It attached no date to any of them.

A month later there are three independent places a shipped Q​wen flagship would be visible, and it is absent from all three. A​libaba Cloud's own Model Studio documentation lists qwen3.8-max as the highest Max tier, with no Q​wen 4 entry. Artificial Analysis, which scores most of the commercial Chinese flagships within days of release, has no Q​wen 4 Max row. The OrcaRouter catalogue, which carries more than 200 models including every Q​wen 3.5 through 3.8 tier, has no qwen4 identifier. That is not proof of a delay. It is proof that as of 22 October there is nothing to compare.

The parameter figure that travelled with the announcement deserves one correction, because it keeps propagating into comparison tables. The 5-trillion-to-10-trillion number in circulation is a forward-looking statement about the generations after Q​wen 4 — the Q​wen 4.5 and Q​wen 5 timeframe. A​libaba has attached no parameter count to Q​wen 4 Max. A Q​wen 5-class size printed next to a Q​wen 4 model name is a borrowing error with a long half-life.

What the announcement does tell you is that A​libaba intends to keep contesting the flagship tier, which is a reason to stay inside the family rather than leave it. It tells you nothing you can plan a migration around, because "soon" is not a date and a conference slide is not a rate card.

Apa yang didapat dari premi spesialis, dan apa yang tidak

GPT-Rosalind's case rests on domain evaluations O​penAI published with the model: leading BixBench performance, six of eleven LABBench2 wins against GPT-5.4, and per-token gains of 53.7% on GeneBench, 18.0% on MedChemBench and 19.6% on LabWorkBench. Every one of those is vendor-reported. None has been reproduced independently, and the model has no entry on the general intelligence boards — which is normal for a gated domain model and also means the numbers cannot be checked from outside.

Yang dapat diperiksa adalah perkakasnya. Life Sciences Research Plugin menghubungkan model ke lebih dari 50 alat dan sumber data ilmiah sebagai keterampilan yang dapat dikomposisikan di dalam Codex, dan makalah 21 Oktober itu adalah demonstrasi tumpukan tersebut, bukan demonstrasi penalaran mentah: ribuan agen yang masing-masing mengambil literatur dan menulis laporan terstruktur merupakan hasil integrasi perkakas. Qwen3.8-Max tidak memiliki tumpukan biologi yang setara. Ia memiliki pemanggilan alat umum dan keluaran terstruktur yang kuat, dan setiap koneksi basis data ilmiah harus Anda bangun sendiri.

Itulah bentuk nyata dari premi tersebut. Anda tidak membayar 2,5× untuk model yang lebih pintar; Anda membayarnya untuk penalaran yang disesuaikan dengan domain atas bukti molekuler ditambah lapisan alat yang sudah terhubung sebelumnya, dan uji coba 21 Oktober adalah bukti publik pertama bahwa kombinasi tersebut menghasilkan sesuatu yang akan dibaca oleh seorang ahli biologi. Kontra yang tidak nyaman ada dalam makalah yang sama: karena eksperimen kontrol mempertahankan model tetap sama dan hanya mengubah dekomposisinya, hasilnya sebagian besar akan tetap bertahan pada model yang lebih murah. Studi tersebut tidak mengisolasi berapa banyak keluaran berasal dari GPT-Rosalind dan berapa banyak dari arsitekturnya — dan para penulis tidak mengklaim demikian. Jika Anda menyusun anggaran untuk penyapuan terdistribusi, ambiguitas itu bernilai sekitar $770 per tiga ribu agen, yang merupakan selisih antara dua tagihan di atas.

Gerbang itu adalah bagian dari spesifikasi

GPT-Rosalind is not generally purchasable. O​penAI's pricing page states that access is limited to approved internal research through the trusted-access programme, and since 11 September that programme has been open globally to eligible organisations rather than to a fixed partner list. Eligibility is assessed on the proposed use, governance and controlled access — which is a reasonable policy for a model that reasons over pathogen and protein data, and a hard constraint on the fleet pattern the company just demonstrated.

Eksekusi selama 32 menit pada ribuan agen mengasumsikan Anda dapat melonjak ke ribuan permintaan bersamaan sesuai permintaan. Pada kunci yang dibatasi aksesnya, kendalanya bukan token per menit; melainkan tinjauan, struktur akun, dan berapa pun batas konkurensi yang diberikan program. Tim mana pun yang merencanakan penyapuan terdistribusi harus memperhitungkan hal itu sebelum membandingkan tarif, karena model yang lebih murah yang dapat Anda panggil dengan bebas mengalahkan model yang lebih baik yang hanya dapat Anda panggil sesuai jadwal orang lain.

This is where a routing layer earns its place, and where it does not. GPT-Rosalind is not on OrcaRouter — access runs through O​penAI's trusted-access application, and no router changes that. What a router does change is everything around the specialist call. A run like the one published on 21 October is mostly not biology: it is orchestration, report parsing, deduplication across thousands of outputs, and the code that turns an atlas into something queryable. Those steps do not need a $25-per-million-output domain model, and putting them there is how a 32-minute experiment becomes an expensive one. Qwen3.8-Max is on OrcaRouter at Alibaba's $2.00/$6.00 list price with 0% markup, behind the same OpenAI-compatible endpoint as more than 200 other models, so the bulk leg and the specialist leg sit on one key with automatic failover between providers and the routing policy written as configuration rather than as a model name baked into the application.

When Q​wen 4 Max ships, that is also the cheapest way to find out whether it matters: a routing rule that sends a slice of traffic to the new tier, with failover back to the model you already trust, is a trial that cannot take production down. Until then the catalogue simply has no entry for it, which is what an announced-and-unreleased model should look like.

Apa yang akan membalikkan perbandingan ini

Four published numbers would turn Q​wen 4 Max from a name into an opponent, and they are worth naming so the real thing is recognisable when it appears.

• The rate card. A Q​wen flagship that undercuts $2.00/$6.00 changes the arithmetic in this article by a factor rather than a percentage. One that matches it changes nothing, because the buyable Q​wen model already sits there.

• Perilaku konteks panjang, dinyatakan sebagai kebijakan. Apakah jendela 1M memiliki pita biaya tambahan, dan pada ambang berapa. Tarif bertahap di atas ukuran input tertentu akan lebih penting bagi sebuah atlas run daripada angka utamanya, karena setiap agen dalam pola itu melintasi pita tersebut.

• An agentic and tool-use result on a board nobody at A​libaba runs. The gap this article describes is not general reasoning — Qwen3.8-Max is competent there — it is scientific tool use. A published, independently administered result on that axis would be the first real test of the specialist premium.

• Whether the Max tier ships weights. Qwen3.8-Max did not, but the 27B tier of the same generation did, and a Q​wen 4 Max with downloadable weights would be the one thing GPT-Rosalind structurally cannot offer a regulated lab that needs to audit what happens to proprietary sequence data.

Keputusan yang sebenarnya didukung halaman ini tidak menunggu salah satu dari mereka. Tempatkan spesialis di tempat kesalahan itu mahal — penentuan prioritas target, proposal mekanisme, laporan yang akan dibaca seorang biolog — dan tempatkan armada pada model yang biayanya $0.14 per agen dan menjawab tanpa formulir pendaftaran. Hasil 21 Oktober adalah argumen yang kuat untuk menjalankan biologi terdistribusi pada skala atlas. Itu argumen yang lebih lemah untuk menjalankan semuanya pada model termahal di ruangan itu, dan eksperimen kontrol dari makalah itu sendiri adalah alasan untuk mengatakan demikian.