Atria Dawn kontra GPT-5.6 Sol: 13-punktowa luka w indeksie i pętla badawcza, która ją ignoruje

Here is the entire tension of this comparison in one line: Atria Dawn Preview is a brand-new open-weights agentic model from the Shanghai AI Laboratory, released on September 14 on a 744B MoE GLM-5.2 base with a 256K context and no published price, while GPT-5.6 Sol is OpenAI​'s flagship tier of the GPT-5.6​ family — shipped July 9, priced at $4.00/$20.00 per million tokens, and carrying an independent Artificial Analysis Intelligence Index of 61 that has been stable through two price cuts. On paper the gap is enormous: a 13-point independent index delta between a closed frontier flagship and an unproven open preview. The reason the matchup is worth reading anyway is that Atria Dawn Preview was not built to close that gap — it was built to make it irrelevant, by targeting the one thing a raw index score does not measure: whether a model will keep working until it has a verifiable result, without a human holding the thread.

Sourcing labels carry this whole article, so they come first. GPT-5.6 Sol's index score, context, and pricing are independent facts you can check on Artificial Analysis and OpenAI​'s own rate page today. Atria Dawn Preview's benchmark rows are vendor-reported from its model card, unreproduced by any neutral lab, and its international API at api.atria-asi.ai has no published price yet. Where the card pits the two against each other, Atria Dawn Preview reports leads on discovery and tool-use suites (BrowseComp 92.5 vs 90.8, BFCL v4 77.0) while GPT-5.6 Sol's coding rows sit far ahead (SWE-bench Pro 59.6 vs a reported 74.7-equivalent on the same grid, Terminal-Bench 2.1 78.3 vs 90.2). None of the Atria numbers is independently verified; treat them as the lab's pitch, not a scorecard.

Two objects that only share the word "model"

GPT-5.6 Sol is OpenAI​'s flagship reasoning tier: a 1.05M-token context window, up to 128K output, text/image/file input, an AA Intelligence Index of 61, and a $4/$20 promotional rate that OpenAI​ cut from $5/$30 on August 24 and has committed to keep at least through November 21 — with a 272K-token cliff above which the entire request bills at $8/$30. It is the model behind Codex, behind Work, and behind the default ChatGPT experience; it is the most widely deployed reasoning model in production right now. It is also proprietary: you rent it, you never own it.

Atria Dawn Preview is the opposite economic object. It is MIT-licensed with downloadable weights (BF16 in 353 shards, FP8 in 177), text-only with a 256K window, deployable through SGLang and vLLM, and hosted internationally at api.atria-asi.ai with the same model id. Its pitch is not raw reasoning — it is the research loop: analyse, design, use tools, write and run code, read the result, recover, iterate. The lab's four pillars — Discovery, Creation, Delivery, Cybersecurity — all reduce to that loop. You cannot download GPT-5.6 Sol; you can download Atria Dawn Preview and run it on hardware you own, and that difference is the entire bet the lab is making.

• Price — Atria Dawn Preview unpublished (international API) vs GPT-5.6 Sol $4.00/$20.00 per 1M ($8/$30 past 272K input)

• Context — Atria 256K text-only vs GPT-5.6 Sol 1.05M, 128K output

• Independent record — none for Atria vs AA Index 61, stable through two repricings

• Openness — Atria MIT weights, self-hostable vs GPT-5.6 Sol proprietary, API-only

What the index gap actually measures — and what it misses

The 13-point index gap is real and you should not hand-wave it. If your workload is a hard reasoning question — a proof, a tricky code bug, a dense legal document — GPT-5.6 Sol will outperform Atria Dawn Preview on average, and the premium is priced accordingly. But the index is built mostly from short, single-shot reasoning tasks, and that is precisely the regime where a research-loop agent is trying to change the rules. Atria Dawn Preview's reported strengths are not short answers; they are BrowseComp, DeepSearchQA, and AutomationBench — long-horizon suites where the model gets to read, act, and correct itself over many turns. The vendor-reported split (wins discovery/tool-use, loses coding) is the signature of a model tuned for iterative work rather than for the index.

For a team that already runs GPT-5.6 Sol for multi-step agentic coding, the honest reading is: Atria Dawn Preview is not a replacement, because its coding rows are exactly where it loses. For a team whose pain is the loop itself — the model stopping to ask, losing the thread, never finishing the experiment — the newcomer is aimed at a real gap, and its open weights mean you can find out for yourself whether the loop holds without paying OpenAI​'s rate for the privilege. Through a routing layer such as OrcaRouter, both models sit behind one key — GPT-5.6 Sol at pass-through list price with automatic failover across providers, and, the moment the Shanghai lab publishes a rate for Atria, the same key routes to it too, so a head-to-head eval costs setup time only.

Werdykt

Do not read the index gap as the whole story, and do not read the vendor grid as proof either — one column is independently verified, the other is a lab's self-report. If you ship software today on GPT-5.6 Sol, nothing here argues for switching; the 13-point delta and the 1.05M window win your workload. If you run research or long-horizon automation where the failure is the model giving up, Atria Dawn Preview's open weights and loop-first tuning are a legitimate experiment worth a controlled eval — on your own hardware, behind failover, without betting a production path on a preview that has not yet been scored by anyone neutral.