Atria Dawn Preview: where it fits

A free, MIT-licensed 744B open-weight model from Shanghai AI Lab that beat GPT-5.6 Sol on the BrowseComp benchmark — released quietly, weights before paper. Released September 11, 2026 by Shanghai AI Laboratory, it is built for best-in-class among tested models on several of Shanghai AI Lab's own benchmarks, including BrowseComp (92.5, ahead of GPT-5.6 Sol's 92.2) and DeepSearchQA (96.0), 744B-parameter MoE built on a GLM-5.2 base, released under a fully open MIT license, and free to self-host — no API pricing, run entirely on your own hardware.

Its trade-offs are real: context window is reported inconsistently across trackers — most list 256K tokens, at least one lists 1M; unconfirmed which is accurate, released quietly, with weights and an API posted to GitHub before Shanghai AI Lab's own technical report went up on arXiv days later, and a lab research preview rather than a commercial product — support and update cadence are unclear. As an open-weight model, its running cost is your own hardware rather than a per-token fee.

Fugu Ultra v2.0: where it fits

Sakana AI's flagship multi-agent orchestration model, claiming best-or-joint-best results on 5 of 8 self-reported benchmarks against monolithic frontier models — $5/$30 per million tokens. Released September 10, 2026 by Sakana AI, it is built for sakana's own evaluation reports best-or-joint-best results on 5 of 8 benchmarks tested, including a 74.3 on the DeepSWE coding-repair benchmark, claims to outperform flagship models like GPT-6 Astra and Claude Fable 5.1 despite its underlying orchestration pool reportedly excluding both of them, and 1M-token context window.

Its trade-offs: not a single trained model — Sakana describes Fugu as "a Multi-Agent System, Delivered as One Model" that dynamically routes tasks across a pool of other models, so head-to-head benchmark framing against monolithic frontier models should be read skeptically, pricing rises to roughly $10/$45 per million tokens for prompts above 272K tokens, and benchmark comparisons against GPT-6 Astra and Claude Fable 5.1 are Sakana's own self-reported figures, not independently verified head-to-head scores. At $5 in / $30 out per million tokens, it sits in the premium price band.

The bottom line for this matchup

The defining split here is open vs. closed. Atria Dawn Preview gives you weights you control — self-host it, fine-tune it, keep data in-house, pay only for hardware. Fugu Ultra v2.0 gives you a managed, always-updated API with no infrastructure to run. Teams with GPUs, privacy requirements, or huge volume often favour the open model; teams that want zero ops and the latest capabilities favour the closed one. Capability is close enough that this operational question, not the benchmark, usually decides it.

Related comparisons