Hy4 Preview: Tencent's 770B-Parameter Open-Weight Model with a 1M Context Window and Binary Reasoning Control

Tencent released Hy4 Preview, an open-weight text-only LLM with 770B total and 49B active parameters, 1M-token context, and 1.56TB of weights on Hugging Face. The chat template restricts reasoning effort to high or no_think, and a sample generation shows clipped hidden reasoning.

August 31, 20261 source75% confidence

63Strategic

Strategic score

Evidence strength38

Technical novelty68

Commercial potential58

Hy4 Preview is Tencent's newest open-weight, text-only LLM, with 770B total parameters, 49B active, a 1M-token context window, and weights totaling 1.56TB on Hugging Face. This more than doubles the total parameter count of Hy3 from July 2026 (295B, 21B active, 256K context) and lengthens the context window by roughly 4x.¹

Context

Two design details stand out. The chat_template.jinja defines reasoning_effort with only two accepted values, 'high' and 'no_think', defaulting to high; no intermediate settings exist. And a single OpenRouter test with the author's standard SVG prompt produced a coherent image, with the exposed reasoning trace written in truncated, telegram-style English.¹

Assessment

The substantive novelty is the binary reasoning switch. Making 'no_think' a first-class mode is an API-contract decision that lets callers trade reasoning for cost and latency without exposing intermediate effort levels. This matters more for serving economics than the raw parameter count.¹

Caveats

AI hypothesisEvidence is limited to a single blog post and one prompt; no benchmark suite or third-party test accompanies the release. The claim that Hy4 is a meaningful long-context open model would be falsified if evaluation shows context degradation near 1M tokens or if the 'no_think' mode removes essential reasoning rather than merely shortening it.¹

Source text

Simon Willison — Introducing Hy4 Preview · August 30, 2026

29th August 2026 - Link Blog

Introducing Hy4 Preview. New open weight text input (no vision) LLM from Chinese company Tencent today: 770B total parameters, 49B active parameters, 1M token context window, 1.56TB on Hugging Face.

This is a big size increase from their previous Hy3 in July, which was 295B, 21B active, 256,000 context, 598GB.

I recently started using model chat templates to better understand their capabilities. Here's Hy4's chat_template.jinja on Hugging Face, which includes this section:

{%- if not reasoning_effort is defined %} {%- set reasoning_effort = 'high' %} {%- elif reasoning_effort not in ['high', 'no_think'] %} {%- if reasoning_effort is none %} {{- raise_exception('reasoning_effort error : None, should be no_think/high') }} {%- else %} {{- raise_exception('reasoning_effort error : ' + reasoning_effort + ', should be no_think/high') }} {%- endif %} {%- endif %}

So it looks like there are just two reasoning effort levels: "high" (the default) and "no_think" (reason by disabled).

I tried my "Generate an SVG of a pelican riding a bicycle" prompt with the default high reasoning via OpenRouter and got this:

Flat vector cartoon illustration of a white pelican with a large orange bill riding a red bicycle to the right along a grey road with a dashed white centre line, its orange webbed feet on the pedals and grey tail feathers fanned out behind, against a pale blue sky with a yellow sun, white clouds and horizontal white motion lines suggesting speed

Quoting the reasoning trace:

[...] Let's maybe add a helmet? It could improve riding theme, but may obscure head. Maybe a small cycling cap or helmet? The user didn't ask; can add red helmet? Might be cute. But pelican with big beak; a helmet might obscure. Better maybe no. Maybe add sunglasses? no. Maybe add water? no.

It's interesting how the reasoning trace uses slightly truncated English, presumably because perfect grammar isn't useful or token efficient for hidden reasoning text.

Context

Assessment

Caveats

Source text

Simon Willison — Introducing Hy4 Preview · August 30, 2026

29th August 2026 - Link Blog

Introducing Hy4 Preview. New open weight text input (no vision) LLM from Chinese company Tencent today: 770B total parameters, 49B active parameters, 1M token context window, 1.56TB on Hugging Face.

This is a big size increase from their previous Hy3 in July, which was 295B, 21B active, 256,000 context, 598GB.

I recently started using model chat templates to better understand their capabilities. Here's Hy4's chat_template.jinja on Hugging Face, which includes this section:

{%- if not reasoning_effort is defined %} {%- set reasoning_effort = 'high' %} {%- elif reasoning_effort not in ['high', 'no_think'] %} {%- if reasoning_effort is none %} {{- raise_exception('reasoning_effort error : None, should be no_think/high') }} {%- else %} {{- raise_exception('reasoning_effort error : ' + reasoning_effort + ', should be no_think/high') }} {%- endif %} {%- endif %}

So it looks like there are just two reasoning effort levels: "high" (the default) and "no_think" (reason by disabled).

I tried my "Generate an SVG of a pelican riding a bicycle" prompt with the default high reasoning via OpenRouter and got this:

Quoting the reasoning trace:

[...] Let's maybe add a helmet? It could improve riding theme, but may obscure head. Maybe a small cycling cap or helmet? The user didn't ask; can add red helmet? Might be cute. But pelican with big beak; a helmet might obscure. Better maybe no. Maybe add sunglasses? no. Maybe add water? no.

It's interesting how the reasoning trace uses slightly truncated English, presumably because perfect grammar isn't useful or token efficient for hidden reasoning text.

Hy4 Preview: Tencent's 770B-Parameter Open-Weight Model with a 1M Context Window and Binary Reasoning Control

Context

Assessment

Caveats

Source text

Recent articles

Simon Willison — Introducing Hy4 Preview

Hy4 Preview: Tencent's 770B-Parameter Open-Weight Model with a 1M Context Window and Binary Reasoning Control

Context

Assessment

Caveats

Source text

Recent articles

Simon Willison — Introducing Hy4 Preview