• bitcoinBitcoin(BTC)$78,418.000.23%
  • ethereumEthereum(ETH)$2,471.171.36%
  • tetherTether(USDT)$1.000.00%
  • binancecoinBNB(BNB)$698.810.87%
  • rippleXRP(XRP)$1.38-4.66%
  • usd-coinUSDC(USDC)$1.000.01%
  • solanaSolana(SOL)$96.940.18%
  • tronTRON(TRX)$0.335784-0.60%
  • Figure HelocFigure Heloc(FIGR_HELOC)$1.00-3.50%
  • HyperliquidHyperliquid(HYPE)$80.471.76%
  • dogecoinDogecoin(DOGE)$0.085305-0.92%
  • zcashZcash(ZEC)$783.621.35%
  • RainRain(RAIN)$0.017487-4.73%
  • USDSUSDS(USDS)$1.00-0.01%
  • whitebitWhiteBIT Coin(WBT)$72.500.46%
  • leo-tokenLEO Token(LEO)$9.25-1.35%
  • chainlinkChainlink(LINK)$11.300.21%
  • moneroMonero(XMR)$432.650.44%
  • cardanoCardano(ADA)$0.205970-1.92%
  • stellarStellar(XLM)$0.179789-2.44%
  • bitcoin-cashBitcoin Cash(BCH)$263.33-0.18%
  • daiDai(DAI)$1.000.01%
  • CantonCanton(CC)$0.115770-2.27%
  • USD1USD1(USD1)$1.000.02%
  • Ethena USDeEthena USDe(USDE)$1.000.01%
  • litecoinLitecoin(LTC)$49.82-0.59%
  • the-open-networkGram (prev. Toncoin)(GRAM)$1.39-5.33%
  • hedera-hashgraphHedera(HBAR)$0.077131-1.37%
  • Global DollarGlobal Dollar(USDG)$1.000.01%
  • avalanche-2Avalanche(AVAX)$7.27-1.24%
  • shiba-inuShiba Inu(SHIB)$0.000005-0.86%
  • suiSui(SUI)$0.74-2.23%
  • Circle USYCCircle USYC(USYC)$1.140.01%
  • crypto-com-chainCronos(CRO)$0.057982-0.62%
  • BlackRock USD Institutional Digital Liquidity FundBlackRock USD Institutional Digital Liquidity Fund(BUIDL)$1.000.00%
  • tether-goldTether Gold(XAUT)$4,584.81-1.26%
  • paypal-usdPayPal USD(PYUSD)$1.000.00%
  • uniswapUniswap(UNI)$4.281.20%
  • MemeCoreMemeCore(M)$1.12-3.42%
  • nearNEAR Protocol(NEAR)$1.84-0.75%
  • okbOKB(OKB)$110.89-2.54%
  • BittensorBittensor(TAO)$228.440.17%
  • Ondo US Dollar YieldOndo US Dollar Yield(USDY)$1.140.11%
  • Ripple USDRipple USD(RLUSD)$1.000.03%
  • pax-goldPAX Gold(PAXG)$4,589.93-1.32%
  • aaveAave(AAVE)$123.45-1.92%
  • AsterAster(ASTER)$0.701.25%
  • World Liberty FinancialWorld Liberty Financial(WLFI)$0.0579410.71%
  • Pump.funPump.fun(PUMP)$0.0046334.50%
  • OndoOndo(ONDO)$0.3634480.56%
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

Alibaba’s Qwen Team Releases Qwen3.8-Flash-Next: A 125B Multimodal MoE With 6B Active Parameters Previewing the Qwen4 Architecture

August 26, 2026
in AI & Technology
Reading Time: 16 mins read
A A
Alibaba’s Qwen Team Releases Qwen3.8-Flash-Next: A 125B Multimodal MoE With 6B Active Parameters Previewing the Qwen4 Architecture
ShareShareShareShareShare

Alibaba’s Qwen team has released Qwen3.8-Flash-Next, an open-weight multimodal Mixture-of-Experts model built for cost per token. The checkpoint pairs a 125B backbone with a 51B N-gram embedding table and a 4B multi-token prediction module. Only 6B parameters activate per token. The team positions it as an early preview of the architecture that will underpin Qwen4, the same role Qwen3-Next played for Qwen3.5. Four changes carry the release: a Gated DeltaNet and Qwen Sparse Attention hybrid, Gated Residual, N-gram Embedding, and the Muon optimizer. Qwen team reports training cost at roughly one-ninth that of Qwen3.7-Plus.

Is it deployable?

Yes but not on a workstation. The FP8 checkpoint is 172.78 GiB and the BF16 checkpoint is 335.28 GiB. Per vLLM recipes, TP2 is the minimum validated FP8 configuration on GB300 and TP4 is recommended. On an 8×H200 node, use TEP8; plain TP8 is incompatible with the checkpoint’s 128-wide quantization blocks. Sparse activation cuts compute, not storage.

What is actually new

Qwen3.8-Flash-Next pairs a 125B main model with 51B N-gram embedding parameters and a 4B multi-token prediction module, totaling 180B on disk. Only 6B parameters activate per token. Four changes drive this:

  • Hybrid attention (GDN + QSA): Three of every four layers use Gated DeltaNet, a linear-attention layer that compresses history into a fixed-size recurrent state. The fourth layer runs Qwen Sparse Attention (QSA), which uses a lightweight indexer to select context at micro-block granularity rather than per token. The layer layout is 12 × (3 × GDN → 1 × QSA) across 48 layers, with a QSA budget of 512 blocks or 2048 tokens.
  • Gated Residual: The residual stream widens into 4 parallel branches, with an element-wise read gate and a per-branch scalar write gate, at bottleneck rank 320.
  • N-gram Embedding: A 20,000,000-entry bigram/trigram table at layer 2 adds capacity through deterministic lookups. It can be offloaded to host memory with asynchronous prefetch — though offload currently runs only on NVIDIA devices.
  • Training recipe.:The Muon optimizer is applied alongside AdamW to specific weight categories, with batch-size warmup eliminated and scaling laws refitted.

The MoE layer carries 512 experts, activating 10 routed plus 1 shared, at expert intermediate dimension 640.

Benchmarks

Qwen reports 58.7 on DeepSWE 1.1, 62.5 on SWE-bench Pro, 81.0 on SWE-bench Multilingual, and 91.9 on LiveCodeBench v6. On agentic tasks it posts 73.9 on CoWorkBench, 55.7 on JobBench, and 73.5 on Toolathlon Verified. Multimodal results include 84.5 on AndroidWorld, 76.6 on LVBench, 88.5 on RealWorldQA, and 95.7 on MathVision with code interpreter.

The model does not lead everywhere. Claude Opus 4.6 (Max) takes HLE at 40.0 against Qwen’s 35.9, and DeepSeek-V4-Flash-0731 leads NL2Repo-Bench at 54.2 versus 48.1. Frontier reasoning remains the gap.

YOU MAY ALSO LIKE

Devolver Digital Squares Up To Rockstar With A Mascot Platformer Out The Same Day As GTA 6

Waystar Puts Agentic AI to Work on Claims, Denials, and Patient Bills – Unite.AI

Efficiency

Qwen states training cost roughly 1/9 that of Qwen3.7-Plus. On serving, the announcement cites QSA kernel speedups of up to 7.6× prefill and 4.9× decode at 1M tokens, while the SGLang cookbook and vLLM recipes cite 10.2× and 6.6×. Treat the range as vendor-reported until independently measured. Qwen also reports 8.6× the prefill throughput of Qwen3.7-Plus at a 90% prefix-cache hit rate.

Context is 262,144 tokens natively, extensible to 1,000,000 with YaRN.

Running it

The model serves through vLLM, SGLang, TokenSpeed, transformers serve, and llama.cpp for GGUF quants. Fine-tuning is supported via Unsloth, Swift, and LLaMA-Factory. It already powers the “Standard” mode on QwenWork and works with Qwen Code.

Thinking mode is on by default, with reasoning_effort at xhigh, medium, or low. Qwen recommends temperature 1.0 and top_p 0.95 for thinking mode, and temperature 0.7 with top_p 0.80 for instruct mode.

Key Takeaways

  • 125B backbone + 51B N-gram embeddings + 4B MTP, with only 6B parameters active per token.
  • Three of four layers use Gated DeltaNet; the fourth runs Qwen Sparse Attention at micro-block granularity.
  • Trained at roughly 1/9 the cost of Qwen3.7-Plus, with 262K native context extensible to 1M via YaRN.
  • FP8 weights are 172.78 GiB, so self-hosting needs a multi-GPU node, not a workstation.
  • Licensed under qwen-community-1.0, not Apache-2.0 — verify terms before commercial use.

Check out the GitHub Page, HF Model Card and Technical Details. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.


Asif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.

Credit: Source link

ShareTweetSendSharePin

Related Posts

Devolver Digital Squares Up To Rockstar With A Mascot Platformer Out The Same Day As GTA 6
AI & Technology

Devolver Digital Squares Up To Rockstar With A Mascot Platformer Out The Same Day As GTA 6

August 26, 2026
Waystar Puts Agentic AI to Work on Claims, Denials, and Patient Bills – Unite.AI
AI & Technology

Waystar Puts Agentic AI to Work on Claims, Denials, and Patient Bills – Unite.AI

August 26, 2026
Orchestration is the new challenge for CX in the age of AI agents
AI & Technology

Orchestration is the new challenge for CX in the age of AI agents

August 26, 2026
What Would Have to Be True for Agentic Coding to Replace Junior Engineers
AI & Technology

What Would Have to Be True for Agentic Coding to Replace Junior Engineers

August 26, 2026
Next Post
Waystar Puts Agentic AI to Work on Claims, Denials, and Patient Bills – Unite.AI

Waystar Puts Agentic AI to Work on Claims, Denials, and Patient Bills – Unite.AI

Leave a Reply Cancel reply

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

Search

No Result
View All Result
OnlyFans model accepts plea deal in boyfriend’s killing

OnlyFans model accepts plea deal in boyfriend’s killing

August 26, 2026
Hundreds evacuated as Indiana flood threat expands

Hundreds evacuated as Indiana flood threat expands

August 22, 2026
Wisconsin’s primary a critical test for Democratic Socialists

Wisconsin’s primary a critical test for Democratic Socialists

August 25, 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!