SShortSingh.
Back to feed

Qwen3-235B-A22B: China's case for large-scale open-weight inference

0
·3 views

Alibaba's Qwen team has released Qwen3-235B-A22B-Instruct-2507, a model with 235B total parameters of which 22B are active per token, reflecting a broader trend toward large sparse open-weight models. The model claims support for 119 languages and up to about one million tokens of extended context, and it is distributed under the permissive Apache 2.0 license, which eases commercial use and redistribution. The article notes that a language list does not replace local evaluation, since regional dialects, domain vocabulary, safety and tool-use quality must be tested before business adoption. Its mixture-of-experts router selects 8 of 128 experts per token, but serving it at full precision still demands significant infrastructure, and long-context use raises costs and information-selection challenges. The author concludes that its practical value depends on measuring quality locally and building the infrastructure and governance needed to run it at scale.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

EU IT tenders and planned US battery capacity: four public-data counts for the week

A DEV Community roundup draws four counts from official public sources. On TED, 884 IT services contract notices were published between 9 and 16 November 2026, with Germany and Poland leading by buyer country. Grants.gov listed only 10 new US federal opportunities posted in the past seven days. CanadaBuys' new-notice file showed 21 notices, and the EIA's September 2026 workbook lists 483 planned battery generator rows totaling 64.2 GW, led by Texas. The author notes that row counts can overstate projects because one plant may span several rows.

0
ProgrammingDEV Community ·

Kiro adds Agent Client Protocol support, letting its agent run inside other IDEs

Kiro has announced support for the Agent Client Protocol in its CLI, allowing its agent capabilities to be used from compatible IDEs and tools rather than only its own editor. The company likens the protocol's potential impact to that of the Language Server Protocol on language support in IDEs. For organizations, this reduces dependence on a single interface and lets teams keep specialized environments, such as those for databases or data science, while accessing the agent. The announcement cites DataGrip as an example, where the IDE handles schema views and connections while the agent assists with queries and migrations. The article stresses that interoperability does not remove the need for identity, authorization, audit trails, and approval rules for high-impact actions.

0
ProgrammingDEV Community ·

Anthropic plans text watermarking for future Claude models to flag AI-generated content

Anthropic has announced that future Claude models will embed watermarks in the text they generate, letting observers estimate the likelihood that a given text was produced at least partly by Claude. The watermark works by choosing alternative, meaning-preserving word options that form a statistical pattern verifiable with a key, and it adds no hidden characters or personal identifiers. It is not a full audit log or a visible label, and it cannot confirm that content is accurate, permitted, or free of human editing. Detection is unreliable for short or factual passages with few linguistic choices, and it does not identify text from other models. Experts advise that such signals should inform risk assessment within existing governance controls rather than trigger automatic decisions.

0
ProgrammingDEV Community ·

Automate Unpaid Invoice Reminders with Make and Gmail

This tutorial shows how to automate follow-ups for overdue invoices using Make, Gmail, and Google Sheets. A Make scenario checks an invoice sheet each morning and sends a gentle reminder at day 3, a firmer one at day 7, and a final notice at day 14 after the due date. Before sending, Make checks Gmail for a client reply and the sheet for a paid status, and skips the reminder if either is found. The guide uses two fictional demo invoices so the setup can be tested safely. It also notes a limitation: payments made outside the sheet will not be detected, so the sheet must be kept current as the source of truth.