Skip to article
AI Pulse
Emergent Story mode

Now reading

Overview

1 / 11 3 min 5 sources Multi-Source
Sources

Story mode

AI PulseMulti-Source5 sections

Liquid AI Ships LFM2.5-230M with llama.cpp, MLX, vLLM, SGLang, and ONNX Support for On-Device Inference

Liquid AI, DeepSeek, Meta, and OpenAI make significant announcements amidst personnel changes

Read
3 min
Sources
5 sources
Domains
2
Sections
5

This week has been marked by significant advancements in AI technology and notable shifts in leadership. Here's a breakdown of the key developments:

Story state
Deep multi-angle story
Evidence
What Happened
Coverage
5 reporting sections
Next focus
What Comes Next

Story step 1

Multi-Source

What Happened

Liquid AI released LFM2.5-230M, its smallest model yet, which runs on-device at 213 tok/s on a Galaxy S25 Ultra and 42 on a Raspberry Pi 5. Built on...

Step
1 / 5

Liquid AI released LFM2.5-230M, its smallest model yet, which runs on-device at 213 tok/s on a Galaxy S25 Ultra and 42 on a Raspberry Pi 5. Built on the LFM2 architecture, it targets tool use and data extraction, beating larger models like Qwen3.5-0.8B and Gemma 3 1B on instruction following.

DeepSeek open-sourced DSpark, a speculative decoding framework that attaches a draft module to existing DeepSeek-V4 weights. It pairs a parallel draft backbone with a lightweight Markov head to cut suffix decay, then adds confidence-scheduled verification that tailors how many tokens get checked to real-time GPU load.

Meta released Astryx, an open-source React design system built on StyleX. It pairs a CSS-variable theme cascade with a CLI and MCP server, so both engineers and AI agents build using the same API.

Meanwhile, Paul Meade, the Apple vice president in charge of the Vision Pro headset, is reportedly leaving the company to join OpenAI’s hardware team.

Continue in the field

Focused storyNearby context

Open the live map from this story.

Carry this article into the map as a focused origin point, then widen into nearby reporting.

Leave the article stream and continue in live map mode with this story pinned as your origin point.

  • Open the map already centered on this story.
  • See what nearby reporting is clustering around the same geography.
  • Jump back to the article whenever you want the original thread.
Open live map mode

Story step 2

Multi-Source

Why It Matters

These developments demonstrate the rapid progress being made in AI research and development. Liquid AI's LFM2.5-230M model showcases the potential...

Step
2 / 5

These developments demonstrate the rapid progress being made in AI research and development. Liquid AI's LFM2.5-230M model showcases the potential for smaller, more efficient models that can run on-device, while DeepSeek's DSpark framework highlights the importance of speculative decoding in accelerating per-user generation.

Meta's Astryx design system, on the other hand, represents a significant step towards creating more accessible and user-friendly AI-powered tools. By providing a common API for both engineers and AI agents, Astryx has the potential to democratize AI development and make it more inclusive.

Story step 3

Multi-Source

Key Numbers

230M: The number of parameters in Liquid AI's LFM2.5-230M model 213 tok/s: The speed at which LFM2.5-230M runs on a Galaxy S25 Ultra 42: The speed at...

Step
3 / 5
  • 230M: The number of parameters in Liquid AI's LFM2.5-230M model
  • 213 tok/s: The speed at which LFM2.5-230M runs on a Galaxy S25 Ultra
  • 42: The speed at which LFM2.5-230M runs on a Raspberry Pi 5
  • 60-85%: The acceleration of DeepSeek-V4 per-user generation over MTP-1 using DSpark

Story step 4

Multi-Source

Key Facts

Who: Liquid AI, DeepSeek, Meta, OpenAI, Apple What: Released new AI models and frameworks, introduced open-source design system, announced personnel...

Step
4 / 5
  • Who: Liquid AI, DeepSeek, Meta, OpenAI, Apple
  • What: Released new AI models and frameworks, introduced open-source design system, announced personnel changes
  • When: This week
  • Where: Global
  • Impact: Significant advancements in AI research and development, potential for more accessible and user-friendly AI-powered tools

Story step 5

Multi-Source

What Comes Next

As AI technology continues to advance at a rapid pace, we can expect to see more innovative applications and use cases emerge. With the release of...

Step
5 / 5

As AI technology continues to advance at a rapid pace, we can expect to see more innovative applications and use cases emerge. With the release of LFM2.5-230M and DSpark, we may see a shift towards more efficient and effective AI models that can run on-device. Meanwhile, Meta's Astryx design system has the potential to democratize AI development and make it more inclusive.

The departure of Paul Meade from Apple to OpenAI may also signal a new era of collaboration and innovation between the two companies. As the AI landscape continues to evolve, one thing is clear: the future of AI is bright, and we can expect to see many more exciting developments in the coming weeks and months.

Cited sources

Multi-Source

5 cited references across 2 linked domains.

References
5
Domains
2

5 cited references across 2 linked domains.

  1. Source 1 · Fulqrum Sources

    Liquid AI Ships LFM2.5-230M with llama.cpp, MLX, vLLM, SGLang, and ONNX Support for On-Device Inference

  2. Source 2 · Fulqrum Sources

    DeepSeek Releases DSpark, a Speculative Decoding Framework That Accelerates DeepSeek-V4 Per-User Generation 60–85% Over MTP-1

  3. Source 3 · Fulqrum Sources

    Apple Vision Pro exec is reportedly leaving for OpenAI

  4. Source 4 · Fulqrum Sources

    Meta’s Astryx Brings a CLI and MCP Server to an Open-Source React Design System Agents Can Read

Open source path

For sponsors

AI PulseDeep read

Reach readers following this story path.

Reach readers choosing AI Pulse coverage with 5 cited references and a clear next-step path.

Evidence
5
Read
3 min

Package the article, desk, and newsletter path around readers already choosing this context.

Sponsor this context

Keep reporting

ContradictionsEvent arcNarrative drift

Open the deeper source boards.

Take the mobile reel into contradictions, event arcs, narrative drift, and the full source workspace.

  • Scan the cited sources and coverage list first.
  • Open contradiction and narrative drift checks after the first read.
  • Revisit the core evidence in What Happened.
Open source boards

Stay in the reporting trail

Open the source boards, cited outlets, and related analysis.

Jump from the app-style read into the deeper source path without losing your place in the story.

Open source pathBack to AI Pulse
🧠 AI Pulse

Liquid AI Ships LFM2.5-230M with llama.cpp, MLX, vLLM, SGLang, and ONNX Support for On-Device Inference

Liquid AI, DeepSeek, Meta, and OpenAI make significant announcements amidst personnel changes

Monday, June 29, 2026 • 3 min read • 5 source references

  • 3 min read
  • 5 source references

This week has been marked by significant advancements in AI technology and notable shifts in leadership. Here's a breakdown of the key developments:

Story pulse
Story state
Deep multi-angle story
Evidence
What Happened
Coverage
5 reporting sections
Next focus
What Comes Next

What Happened

Liquid AI released LFM2.5-230M, its smallest model yet, which runs on-device at 213 tok/s on a Galaxy S25 Ultra and 42 on a Raspberry Pi 5. Built on the LFM2 architecture, it targets tool use and data extraction, beating larger models like Qwen3.5-0.8B and Gemma 3 1B on instruction following.

DeepSeek open-sourced DSpark, a speculative decoding framework that attaches a draft module to existing DeepSeek-V4 weights. It pairs a parallel draft backbone with a lightweight Markov head to cut suffix decay, then adds confidence-scheduled verification that tailors how many tokens get checked to real-time GPU load.

Meta released Astryx, an open-source React design system built on StyleX. It pairs a CSS-variable theme cascade with a CLI and MCP server, so both engineers and AI agents build using the same API.

Meanwhile, Paul Meade, the Apple vice president in charge of the Vision Pro headset, is reportedly leaving the company to join OpenAI’s hardware team.

Advertisement

Ad slot: in-article

Why It Matters

These developments demonstrate the rapid progress being made in AI research and development. Liquid AI's LFM2.5-230M model showcases the potential for smaller, more efficient models that can run on-device, while DeepSeek's DSpark framework highlights the importance of speculative decoding in accelerating per-user generation.

Meta's Astryx design system, on the other hand, represents a significant step towards creating more accessible and user-friendly AI-powered tools. By providing a common API for both engineers and AI agents, Astryx has the potential to democratize AI development and make it more inclusive.

Key Numbers

  • 230M: The number of parameters in Liquid AI's LFM2.5-230M model
  • 213 tok/s: The speed at which LFM2.5-230M runs on a Galaxy S25 Ultra
  • 42: The speed at which LFM2.5-230M runs on a Raspberry Pi 5
  • 60-85%: The acceleration of DeepSeek-V4 per-user generation over MTP-1 using DSpark

Key Facts

  • Who: Liquid AI, DeepSeek, Meta, OpenAI, Apple
  • What: Released new AI models and frameworks, introduced open-source design system, announced personnel changes
  • When: This week
  • Where: Global
  • Impact: Significant advancements in AI research and development, potential for more accessible and user-friendly AI-powered tools

What Comes Next

As AI technology continues to advance at a rapid pace, we can expect to see more innovative applications and use cases emerge. With the release of LFM2.5-230M and DSpark, we may see a shift towards more efficient and effective AI models that can run on-device. Meanwhile, Meta's Astryx design system has the potential to democratize AI development and make it more inclusive.

The departure of Paul Meade from Apple to OpenAI may also signal a new era of collaboration and innovation between the two companies. As the AI landscape continues to evolve, one thing is clear: the future of AI is bright, and we can expect to see many more exciting developments in the coming weeks and months.

Coverage tools

Sources, context, and related analysis

Source path

How this briefing, its cited outlets, and the next reporting move fit together

A compact source board that keeps the article legible while showing what supports the current read and what would most improve the coverage next.

Cited sources

0

Reading points

3

Source links

2

Next checks

1

Source map

From briefing to cited outlets to next reporting move

Source path ready

Story geography

Where this reporting sits on the map

Use the map-native view to understand what is happening near this story and what adjacent reporting is clustering around the same geography.

Geo context
0.00° N · 0.00° E Mapped story

This story is geotagged. Nearby related reporting is not ready yet, so the live map is the best next context check.

Continue in live map mode

Coverage at a Glance

5 sources

Compare coverage, inspect perspective spread, and open primary references side by side.

Linked Sources

5

Distinct Outlets

2

Viewpoint Center

Center

Outlet Diversity

Very Narrow
2 sources with viewpoint mapping 2 higher-credibility sources

Coverage Gaps to Watch

No major coverage gaps detected in the current source set. Recheck as new reporting comes in.

Read Across More Angles

Source-by-Source View

Search by outlet or domain, then filter by credibility, viewpoint mapping, or the most-cited lane.

Showing 5 of 5 cited sources with links.

Center (2)

TechCrunch

Apple Vision Pro exec is reportedly leaving for OpenAI

Open

techcrunch.com

Center High Dossier
TechCrunch

Trump Admin releases Anthropic Mythos to be used by more than 100 US companies, agencies

Open

techcrunch.com

Center High Dossier

Unmapped Perspective (3)

marktechpost.com

Liquid AI Ships LFM2.5-230M with llama.cpp, MLX, vLLM, SGLang, and ONNX Support for On-Device Inference

Open

marktechpost.com

Unmapped bias Credibility unknown Dossier
marktechpost.com

DeepSeek Releases DSpark, a Speculative Decoding Framework That Accelerates DeepSeek-V4 Per-User Generation 60–85% Over MTP-1

Open

marktechpost.com

Unmapped bias Credibility unknown Dossier
marktechpost.com

Meta’s Astryx Brings a CLI and MCP Server to an Open-Source React Design System Agents Can Read

Open

marktechpost.com

Unmapped bias Credibility unknown Dossier
Source-linked Fast briefing Contrast-aware

Emergent News uses automated assistance to gather, compare, and summarize coverage from 5 cited sources. Review the source list below before relying on the story.