• bitcoinBitcoin(BTC)$77,684.00-0.53%
  • ethereumEthereum(ETH)$2,318.20-0.62%
  • tetherTether(USDT)$1.00-0.01%
  • rippleXRP(XRP)$1.41-0.99%
  • binancecoinBNB(BNB)$626.15-0.97%
  • usd-coinUSDC(USDC)$1.00-0.01%
  • solanaSolana(SOL)$85.31-1.56%
  • tronTRON(TRX)$0.323671-0.10%
  • Figure HelocFigure Heloc(FIGR_HELOC)$1.020.00%
  • dogecoinDogecoin(DOGE)$0.097803-0.73%
  • whitebitWhiteBIT Coin(WBT)$54.84-0.75%
  • USDSUSDS(USDS)$1.000.01%
  • HyperliquidHyperliquid(HYPE)$42.523.00%
  • leo-tokenLEO Token(LEO)$10.371.04%
  • cardanoCardano(ADA)$0.247097-1.74%
  • bitcoin-cashBitcoin Cash(BCH)$447.21-1.49%
  • moneroMonero(XMR)$388.761.10%
  • chainlinkChainlink(LINK)$9.32-1.07%
  • zcashZcash(ZEC)$357.390.27%
  • CantonCanton(CC)$0.149295-0.57%
  • stellarStellar(XLM)$0.168147-1.58%
  • MemeCoreMemeCore(M)$4.22-2.73%
  • daiDai(DAI)$1.000.01%
  • USD1USD1(USD1)$1.00-0.02%
  • litecoinLitecoin(LTC)$55.29-1.77%
  • avalanche-2Avalanche(AVAX)$9.24-2.06%
  • hedera-hashgraphHedera(HBAR)$0.090680-1.72%
  • Ethena USDeEthena USDe(USDE)$1.00-0.01%
  • suiSui(SUI)$0.93-2.29%
  • shiba-inuShiba Inu(SHIB)$0.000006-1.70%
  • RainRain(RAIN)$0.0074524.56%
  • paypal-usdPayPal USD(PYUSD)$1.000.01%
  • the-open-networkToncoin(TON)$1.30-1.42%
  • crypto-com-chainCronos(CRO)$0.069865-0.76%
  • Circle USYCCircle USYC(USYC)$1.120.00%
  • tether-goldTether Gold(XAUT)$4,694.760.03%
  • Global DollarGlobal Dollar(USDG)$1.00-0.02%
  • BittensorBittensor(TAO)$248.700.62%
  • World Liberty FinancialWorld Liberty Financial(WLFI)$0.072974-2.88%
  • BlackRock USD Institutional Digital Liquidity FundBlackRock USD Institutional Digital Liquidity Fund(BUIDL)$1.000.00%
  • pax-goldPAX Gold(PAXG)$4,700.320.15%
  • mantleMantle(MNT)$0.64-2.43%
  • polkadotPolkadot(DOT)$1.24-2.06%
  • uniswapUniswap(UNI)$3.23-1.55%
  • SkySky(SKY)$0.084639-3.92%
  • Pi NetworkPi Network(PI)$0.1775930.07%
  • Falcon USDFalcon USD(USDF)$1.00-0.02%
  • nearNEAR Protocol(NEAR)$1.37-2.64%
  • okbOKB(OKB)$83.84-0.57%
  • HTX DAOHTX DAO(HTX)$0.000002-0.31%
TradePoint.io
  • Main
  • AI & Technology
  • Stock Charts
  • Market & News
  • Business
  • Finance Tips
  • Trade Tube
  • Blog
  • Shop
No Result
View All Result
TradePoint.io
No Result
View All Result

Liquid AI Released LFM2.5-350M: A Compact 350M Parameter Model Trained on 28T Tokens with Scaled Reinforcement Learning

April 1, 2026
in AI & Technology
Reading Time: 6 mins read
A A
Liquid AI Released LFM2.5-350M: A Compact 350M Parameter Model Trained on 28T Tokens with Scaled Reinforcement Learning
ShareShareShareShareShare

In the current landscape of generative AI, the ‘scaling laws’ have generally dictated that more parameters equal more intelligence. However, Liquid AI is challenging this convention with the release of LFM2.5-350M. This model is actually a technical case study in intelligence density with additional pre-training (from 10T to 28T tokens) and large-scale reinforcement learning

The significance of LFM2.5-350M lies in its architecture and training efficiency. While the most AI companies has been focused on frontier models, Liquid AI is targeting the ‘edge’—devices with limited memory and compute—by proving that a 350-million parameter model can outperform models more than twice its size on several evaluated benchmarks.

YOU MAY ALSO LIKE

The LoRA Assumption That Breaks in Production 

How to Build Smarter Multilingual Text Wrapping with BudouX Through Parsing, HTML Rendering, Model Introspection, and Toy Training

https://www.liquid.ai/blog/lfm2-5-350m-no-size-left-behind

Architecture: The Hybrid LIV Backbone

The core technical differentiator of the LFM2.5-350M is its departure from the pure Transformer architecture. It utilizes a hybrid structure built on Linear Input-Varying Systems (LIVs).

Traditional Transformers rely entirely on self-attention mechanisms, which suffer from quadratic scaling issues: as the context window grows, the memory and computational requirements for the Key-Value (KV) cache increase. Liquid AI addresses this by using a hybrid backbone consisting of:

  • 10 Double-Gated LIV Convolution Blocks: These handle the majority of the sequence processing. LIVs function similarly to advanced Recurrent Neural Networks (RNNs) but are designed to be more parallelizable and stable during training. They maintain a constant-state memory, reducing the I/O overhead.
  • 6 Grouped Query Attention (GQA) Blocks: By integrating a small number of attention blocks, the model retains high-precision retrieval and long-range context handling without the full memory overhead of a standard Transformer.

This hybrid approach allows the LFM2.5-350M to support a 32k context window (32,768 tokens) while maintaining an extremely lean memory footprint.

Performance and Intelligence Density

The LFM2.5-350M was pre-trained on 28 trillion tokens with an extremely high training-to-parameter ratio. This ensures that the model’s limited parameter count is utilized to its maximum potential, resulting in high ‘intelligence density.’

Benchmarks and Use Cases

The LFM2.5-350M is a specialist model designed for high-speed, agentic tasks rather than general-purpose reasoning.

Benchmark Score
IFEval (Instruction Following) 76.96
GPQA Diamond 30.64
MMLU-Pro 20.01

The high IFEval score indicates the model is efficient at following complex, structured instructions, making it suitable for tool use, function calling, and structured data extraction (e.g., JSON). However, the documentation explicitly states that LFM2.5-350M is not recommended for mathematics, complex coding, or creative writing. For those tasks, the reasoning capabilities of larger parameter counts remain necessary.

https://www.liquid.ai/blog/lfm2-5-350m-no-size-left-behind

Hardware Optimization and Inference Efficiency

A major hurdle for AI devs is the ‘memory wall’—the bottleneck created by moving data between the processor and memory. Because the LFM2.5-350M utilizes LIVs and GQA, it drastically reduces KV cache size, boosting throughput. On a single NVIDIA H100 GPU, the model can reach a throughput of 40.4K output tokens per second at high concurrency.

Liquid AI team reports device-specific low-memory inference results that make local deployment viable:

  • Snapdragon 8 Elite NPU: 169MB peak memory using RunAnywhere Q4.
  • Snapdragon GPU: 81MB peak memory using RunAnywhere Q4.
  • Raspberry Pi 5: 300MB using Cactus Engine int8.

Key Takeaways

  • Extreme Intelligence Density: By training a 350M parameter model on 28 trillion tokens, Liquid AI team achieved an super high 80,000:1 token-to-parameter ratio, allowing it to outperform models more than twice its size on several benchmarks.
  • Hybrid LIV Architecture: The model departs from pure Transformers by using Linear Input-Varying Systems (LIVs) combined with a small number of Grouped Query Attention (GQA) blocks, significantly reducing the memory overhead of the KV cache.
  • Edge-First Efficiency: It is designed for local deployment with a 32k context window and a remarkably low memory footprint—reaching as low as 81MB on mobile GPUs and 169MB on NPUs via specialized inference engines.
  • Specialized Agentic Capability: The model is highly optimized for instruction following (IFEval: 76.96) and tool use, though it is explicitly not recommended for complex coding, mathematics, or creative writing.
  • Massive Throughput: The architectural efficiency enables high-speed utility, processing up to 40.4K output tokens per second on a single H100, making it ideal for high-volume data extraction and real-time classification.

Check out the Technical details and Model Weight. Also, feel free to follow us on Twitter and don’t forget to join our 120k+ ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.

The post Liquid AI Released LFM2.5-350M: A Compact 350M Parameter Model Trained on 28T Tokens with Scaled Reinforcement Learning appeared first on MarkTechPost.

Credit: Source link

ShareTweetSendSharePin

Related Posts

The LoRA Assumption That Breaks in Production 
AI & Technology

The LoRA Assumption That Breaks in Production 

April 27, 2026
How to Build Smarter Multilingual Text Wrapping with BudouX Through Parsing, HTML Rendering, Model Introspection, and Toy Training
AI & Technology

How to Build Smarter Multilingual Text Wrapping with BudouX Through Parsing, HTML Rendering, Model Introspection, and Toy Training

April 26, 2026
Forced Windows updates can now be paused forever
AI & Technology

Forced Windows updates can now be paused forever

April 26, 2026
Canadian premier wants to ban social media and AI chatbots for kids in Manitoba
AI & Technology

Canadian premier wants to ban social media and AI chatbots for kids in Manitoba

April 26, 2026
Next Post
Austin police release bodycam video from deadly bar shooting

Austin police release bodycam video from deadly bar shooting

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Search

No Result
View All Result
Video captures the moment a toddler in Arizona was rescued from oncoming traffic

Video captures the moment a toddler in Arizona was rescued from oncoming traffic

April 23, 2026
7.7-magnitude earthquake strikes off Japan’s coast

7.7-magnitude earthquake strikes off Japan’s coast

April 21, 2026
Extended interview: Sen. Tillis on not running for re-election and Trump’s Cabinet

Extended interview: Sen. Tillis on not running for re-election and Trump’s Cabinet

April 24, 2026

About

Learn more

Our Services

Legal

Privacy Policy

Terms of Use

Bloggers

Learn more

Article Links

Contact

Advertise

Ask us anything

©2020- TradePoint.io - All rights reserved!

Tradepoint.io, being just a publishing and technology platform, is not a registered broker-dealer or investment adviser. So we do not provide investment advice. Rather, brokerage services are provided to clients of Tradepoint.io by independent SEC-registered broker-dealers and members of FINRA/SIPC. Every form of investing carries some risk and past performance is not a guarantee of future results. “Tradepoint.io“, “Instant Investing” and “My Trading Tools” are registered trademarks of Apperbuild, LLC.

This website is operated by Apperbuild, LLC. We have no link to any brokerage firm and we do not provide investment advice. Every information and resource we provide is solely for the education of our readers. © 2020 Apperbuild, LLC. All rights reserved.

No Result
View All Result
  • Main
  • AI & Technology
  • Stock Charts
  • Market & News
  • Business
  • Finance Tips
  • Trade Tube
  • Blog
  • Shop

© 2023 - TradePoint.io - All Rights Reserved!