Findings from the data.
What changed, who should take another look, and the evidence behind it.
The Glimmer Report
We check the coding and document guides every week. A report follows when the leading candidates change, including open-weight options. New tests can matter even when no model launches. Quiet weeks have no edition.
Last checked Oct 26, 2026. No change in the leading candidates. No report this week.
Choosing a model today? Start with a living guide: coding or document extraction.
Release report · Oct 26, 2026GPT-6.1 Terra: release reportOpenAI released GPT-6.1 Terra on October 26, 2026. Specs, price, how it plays chess and how it solves sudoku, with more added as evidence arrives.Release report · Oct 20, 2026Voxtral 3: release reportMistral released Voxtral 3 on October 20, 2026. Specs, with more added as evidence arrives.The Glimmer Report · Oct 19, 2026A fresh model comparison for coding and document workThe leading candidates changed for coding and document work. Here's what moved and how to compare it with your current setup.Release report · Oct 17, 2026DeepSeek-V4.1-Flash-1017: release reportDeepSeek released DeepSeek-V4.1-Flash-1017 on October 17, 2026. Specs and price, with more added as evidence arrives.Release report · Oct 14, 2026Grok Voice 3.0: release reportxAI released Grok Voice 3.0 on October 14, 2026. Specs, with more added as evidence arrives.The Glimmer Report · Oct 12, 2026A fresh model comparison for codingThe leading candidates changed for coding. Here's what moved and how to compare it with your current setup.Release report · Oct 7, 2026Claude Haiku 5.5: release reportAnthropic released Claude Haiku 5.5 on October 7, 2026. Specs, price, where it ranks, how it plays chess and how it solves sudoku.Release report · Oct 6, 2026Mistral Large 4: release reportMistral released Mistral Large 4 on October 6, 2026. Specs, price and where it ranks, with more added as evidence arrives.Finding · Sep 15, 2026Open weights have trailed the frontier by four to eight months since spring 2025Every time an open-weights model set a new best score since January 2025, the closed frontier had already been there for between 1 and 8 months. Kimi K3 brought the gap to 4.4 months, the low end of the range, not a new regime.