Latest

The 300x spread in cost per task means the ranking alone is misleading: GPT-6.1 Sol delivers fourth place at 90% less cost than the leader, which changes which model is actually practical for batch video generation. Teams building automated marketing video pipelines can use the five-axis Elo breakdown to match models to their budget and quality needs, and the open-weights results give a concrete quality gap for self-hosting comparisons.

Official accountArtificial Analysis@ArtificialAnlysSource

Open weights make private genomic analysis possible at a fraction of frontier-model cost, which matters for labs handling sensitive sequencing data that cannot leave their infrastructure. The bundled Claude Code and Codex Skills mean bioinformatics teams can plug DNA sequence analysis into existing agent pipelines without building custom integrations. The Qwen3 base also gives practitioners familiar with the Qwen ecosystem a low-friction path to fine-tuning.

AIGCLINK@aigclinkSource

The pricing is the notable part: at two to three cents per task, Nemotron 3.5 Super VL is positioned as a low-cost vision-language option even though its abstract-reasoning scores trail the leaders. Teams benchmarking multimodal models against ARC-AGI can compare cost-performance trade-offs directly, since ARC Prize reports price alongside accuracy for every entry.

Official accountARC Prize@arcprizeSource

Sakana's shift signals that agent-first labs are re-entering base-model development, betting that small teams with AI-managed experiment pipelines can compete on limited compute. Japanese-language post-training (Namazu) and orchestration (Fugu) offer a template for non-English markets building sovereign AI stacks. Teams weighing whether to build or fine-tune models can benchmark Sakana's three-pillar framing against single-model strategies.

Official accountSakana AI@SakanaAILabsSource

The simple Llama-style architecture means almost any inference engine can run it out of the box, at the cost of high memory usage and no built-in speculative decoding beyond external MTP. Agent-focused teams looking for an open teacher model to distil from should benchmark it against Qwen3.8-27B now, and watch for the promised sub-10B distilled releases. BOSS Zhipin being a recruiter entering open-weight modelling is itself notable for the China AI ecosystem.

karminski-牙医 (AI工具)@karminski3Source

The non-obvious shift here is that a frontier Chinese model can now be customised inside a US platform, with fine-tuning runs staying in the customer's own Python loop rather than a black-box pipeline. Teams building agents can adapt M3.1 Flash with custom reward signals and environments via SFT or RL. This makes MiniMax a direct alternative to open-weight models fine-tuned on Together, with the model served alongside Western options.

Official accountTogether AI@togethercomputeSource