Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Tencent launches Hy4 Flash with a 1M-token context at $0.07 per million tokens

Tencent has released Hy4 Flash, a model with a 1-million-token context window priced at $0.07 per million tokens, according to a report by Pasquale Pillitteri. The pairing of very long context and very low unit cost targets repository-scale code work, long-document analysis and persistent agent sessions, though independent benchmarks and availability details are still missing.

Published
腾讯推出 Hy4 Flash:100 万 token 上下文,每百万 token 0.07 美元
Image source: tencent.com

Tencent has released a model called Hy4 Flash, according to a report by Pasquale Pillitteri, and two numbers carry most of the story: a 1-million-token context window and a price of $0.07 per million tokens.

A million tokens is roughly several hundred pages of text, or the bulk of a mid-sized codebase. At that price, loading so much material into a single request stops being a cost that needs justifying and becomes something a developer can do by default. At that price, the more consequential change is not what the model can do but which applications suddenly become cheap enough to run.

The combination points at a familiar set of workloads: searching and rewriting an entire repository, comparing long contracts or financial filings line by line, multi-turn agent sessions that must hold state for a long time, and bulk structured extraction from document sets. Such tasks have typically required chunking, embedding and retrieval before a model could touch them; when context is both long and cheap, some of that engineering can be skipped.

The report leaves a lot unanswered. It gives no parameter count, architecture or indication of whether weights will be released, and it does not say whether the $0.07 applies to input or output tokens, or whether caching and batch discounts change the figure. Nor does it cover retrieval quality at different positions in a long context, rate limits or which regions can access the model — the details that decide whether it works in production rather than merely looking good in an announcement.

Set against the wider market, long context and low prices have become the two main axes of competition in model APIs this year, and each order-of-magnitude cut in unit cost makes a class of applications viable that previously was not. Demand for whole-repository code review or large-scale contract analysis has never been the missing piece; the bill usually was.

Three things are worth watching: independent benchmarks for long-context recall, coding and tool use; real throughput and rate-limit policy; and whether Tencent publishes weights or widens regional availability. The numbers are on the table now, and those three will decide whether they hold up.

Why it matters

If the pricing holds in practice, long context shifts from an occasional feature to a default setting, with the biggest effect on document- and code-heavy workflows. Without architecture, rate-limit and benchmark detail, it reads as a pricing signal rather than a verified capability claim.

TencentHy4 FlashAPI PricingLong Context
Back to realtime news

Nearby Updates

All

11/17, 01:02

OpenAI embeds Apollo Research evaluators to probe whether its models try to subvert oversight

OpenAI will station six Apollo Research safety evaluators inside the company as a pilot, investigating whether its models attempt to undermine the alignment and control mechanisms that keep them, or their successors, in check. OpenAI is granting scoped access to training code, datasets, reinforcement-learning rollouts and intermediate checkpoints while reserving the right to withhold material on security or IP grounds, leaving the pilot's independence in question.

11/17, 01:00

OpenAI ships a new image classifier and 24-hour ChatGPT bans to fight CSAM

OpenAI released a new image classifier aimed at child sexual abuse material and changed how ChatGPT handles accounts that seek it out, TechCrunch reported. Under the new behaviour, an account trying to access CSAM content receives a warning and a 24-hour ban.

11/17, 00:36

Qwen4-Max goes official with 2.4T MoE parameters and a claimed 60% cost cut

The official version of Alibaba's Qwen4-Max is live, with Sohu reporting a 2.4-trillion-parameter mixture-of-experts architecture and task costs roughly 60% lower. Related coverage also cites a 1-million-token context window for long documents and complex workloads.

11/17, 00:07

Report: OpenAI Explored Buying Interpretability Startup Goodfire for Over $5 Billion

OpenAI explored acquiring the AI interpretability startup Goodfire in a transaction valued at more than $5 billion, The Information reported. The figure, if accurate, would rank among the largest sums ever attached to a company working on AI interpretability.