Question
Will any Thinking Machines Lab model rank in the top 10 of the Artificial Analysis Intelligence Index at any point before January 1, 2028?
Current Standing and the Capability Gap Thinking Machines Lab (TML) faces a formidable climb to reach the frontier. As of November 2026, TML's strongest model, Inkling 2 Small, scores 30.6 on version 5.0.1 of the Artificial Analysis Intelligence Index. The 10th-place entry, GPT-6 Astra (xhigh), sits at 55.3 artificialanalysis.ai. Bridging a nearly 25-point gap requires a massive leap in capabilities, especially against an accelerating frontier. While TML released the model's weights under an Apache 2.0 license, making it the highest-scoring open-weights model outside China, this status does not improve its overall index rank.
The Structural Hurdle of the Top 10 The top 10 is an exceptionally difficult target because Artificial Analysis evaluates reasoning-effort variants as separate entries artificialanalysis.ai. Anthropic’s Claude Fable 5.2 alone occupies three of the top five slots. With variants of Gemini 4 Carbon, Claude Opus 5.5, and GPT-6 Astra filling out the list, the top 10 is effectively locked down by the three leading labs. To qualify, TML would not just need to become a top-10 AI company; it would have to outcompete multiple model variants from the leaders, as well as leapfrog far better-resourced challengers like Meta (50.1), Moonshot (46.5), and xAI (35.5), none of which are currently in the top 10.
Resources and the Tight 2027 Timeline There are credible upside paths for TML. The company explicitly plans a "ladder of models" with larger releases on the horizon thinkingmachines.ai, backed by reported advanced talks for a $5 billion funding round at a $40 billion valuation businessinsider.com. Furthermore, TML has a gigawatt-scale Nvidia Vera Rubin deployment targeted for early 2027 blogs.nvidia.com. However, this timeline leaves little room for error. Siting and commissioning gigawatt-scale infrastructure, completing a massive training run, and conducting post-training will realistically push any frontier-class release to late 2027. Furthermore, scaling is no guarantee of success: TML's earlier 41B-parameter Inkling scored no better than its 12B-parameter Inkling-Small 2 sources.
A Moving Target and Emerging Headwinds Even if TML successfully executes a massive scale-up in late 2027, the 10th-place threshold will not remain static. Evidence suggests that AI research is accelerating, with internal progress at Anthropic running at roughly twice its historical rate. Incumbents are also taking steps to secure their leads; for example, Anthropic has introduced R&D classifiers and defenses that drastically reduce the data available for smaller labs to distill. Combined with significant talent bleed—four of TML's six original co-founders have departed—the odds that TML can close a 25-point gap against a rapidly advancing frontier before January 2028 are slim.
The estimate was maintained at 5%, remaining consistent with the high difficulty of displacing incumbent frontier models.
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited