▲Frontier Model Efficiency Gains

Score history

Daily conviction score, 0 to 100. Higher means the thesis is more strongly corroborated.

Now 98 · +3 since Oct 9 · ranged 95 to 98

Showing the last few days. Unlock full score history.

Why the conviction moved

• Oct 10
  Strengthened +3

  Haiku 5.5 raises its OSWorld score to 72.4% at one-tenth of its predecessor's price, with a one-million-token context and adjustable effort. That is more capability per dollar at Anthropic's smallest tier.

• Oct 8
  Strengthened +2

  Google reports that EmbeddingGemma 2 adds nearly 10 points over the first version on the Massive Text Embedding Benchmark's code track while still fitting in phone memory. That is more capability per parameter at sub-billion scale.

Showing the last 2 days. Unlock the full record.

Source trail

• Supporting · October 10, 2026

  Anthropic prices Claude Haiku 5.5 at one-tenth of Haiku 4.5 and lifts its OSWorld score to 72.4%

  Haiku 5.5 raises its OSWorld score to 72.4% at one-tenth of its predecessor's price, with a one-million-token context and adjustable effort. That is more capability per dollar at Anthropic's smallest tier.

  Anthropic News

• Supporting · October 8, 2026

  Google releases EmbeddingGemma 2, a 740-million-parameter open embedder for text, code, images and audio under Apache 2.0

  Google reports that EmbeddingGemma 2 adds nearly 10 points over the first version on the Massive Text Embedding Benchmark's code track while still fitting in phone memory. That is more capability per parameter at sub-billion scale.

  Google DeepMind