Skip to content
Drix10 Blog

Meta Quest 2 & Pro final OS v2.13 March 2027

, 5 items in AR VR Professionals and Community, 3 min read

In this digest (5 items)

Meta Quest 2 and Meta Quest Pro will receive their final OS release v2.13, scheduled for all devices by March 2027. Existing apps keep working and can still receive updates.

Key points

  • Final OS release: v2.13

  • Rollout date: March 2027 for all devices

  • Existing apps remain functional and can receive updates

Sources

Weight-Space Programming edits 0.016% of Qwen3.5-0.8B weights, boosting math 6×

Weight-Space Programming used GPT-6 Astra to edit Qwen3.5-0.8B weights, changing only 0.016% of parameters. The edit fixed reasoning loops and improved math performance up to 6× and instruction following up to 3×, without gradients or training.

Key points

  • Changed 0.016% of Qwen3.5-0.8B weights.

  • Math performance improved up to 6×; instruction following improved up to 3×.

Sources

Grok Bot live dashboard setup with single prompt

Grok Bot can create a live dashboard for agent teams using a single prompt that consumes 2% of usage. The prompt makes the bot manage a private office channel, adds status rules, and displays roster and team locations. Open the provided link to view.

Key points

  • Prompt creates a private channel that only you can read

  • Prompt adds one status rule to every bot, removable with one message

Sources

Meta Quest 2 and Quest Pro final OS v2.13 by March 2027

Meta Quest 2 and Meta Quest Pro will receive their final OS release, version v2.13, which will be delivered to all devices by March 2027. Existing apps will continue to work and can still receive updates.

Key points

  • Final OS release: v2.13

  • Delivery deadline: March 2027 for all devices

  • Existing apps remain functional and updatable

Sources

Hume Audio Decision API launch, 10M+ judgments, one call replaces 10+ models

Hume Audio Decision API lets you ask questions such as “How many speakers?”, “Is anyone laughing?” or “Is the caller frustrated?” It is trained on over 10 million human judgments and consolidates what used to require ten or more models into a single call.

Key points

  • Trained on 10M+ human judgments.

  • One call replaces 10+ models.

Sources