=== HEADLINE_SHORT ===
Claude Just Got Tied. By Open Source.

=== HOOK_SCRIPT ===
A Chinese AI lab just tied Claude Opus on the hardest coding benchmark in the industry.

=== SHORT_SCRIPT ===
A Chinese AI lab just tied Claude Opus on the hardest coding benchmark in the industry. Then they released the weights for FREE. And priced the API at one seventh of what Anthropic charges.

Here is the number. Anthropic's Claude Opus 4.6 sits at 80.8 percent on SWE-bench Verified. That is the test that asks a model to actually fix real GitHub issues end to end. It is the benchmark every coding agent company has been chasing for a year. DeepSeek V4-Pro just hit 80.6 percent. Two tenths of one percentage point. And you can download the weights from Hugging Face right now, MIT licensed, run it on your own hardware, fine tune it however you want.

V4-Pro also leads on LiveCodeBench, on Codeforces, on Terminal-Bench 2.0. Three out of four headline coding evals. The closed labs still win on world knowledge and the hardest reasoning. But coding is the use case paying the bills right now, and on coding the gap is GONE.

This is the floor. Open weights at frontier. From here the only direction is down on price and up on capability. Stay sharp.

=== ANNOTATED_HTML_SHORT ===
<span class="cue">[TIGHT CLOSE-UP — Jane]</span>
<p>A Chinese AI lab just tied Claude Opus on the hardest coding benchmark in the industry.</p>
<span class="cue">[ON-SCREEN TEXT: "80.6% SWE-Bench"]</span>
<p>Then they released the weights for FREE. And priced the API at one seventh of what Anthropic charges.</p>
<span class="cue">[ZOOM IN]</span>
<p>Here is the number. Anthropic's Claude Opus 4.6 sits at 80.8 percent on SWE-bench Verified. That is the test that asks a model to actually fix real GitHub issues end to end. It is the benchmark every coding agent company has been chasing for a year. DeepSeek V4-Pro just hit 80.6 percent. Two tenths of one percentage point. And you can download the weights from Hugging Face right now, MIT licensed, run it on your own hardware, fine tune it however you want.</p>
<span class="cue">[ON-SCREEN TEXT: "$1.74 / $3.48 per M tokens"]</span>
<p>V4-Pro also leads on LiveCodeBench, on Codeforces, on Terminal-Bench 2.0. Three out of four headline coding evals. The closed labs still win on world knowledge and the hardest reasoning. But coding is the use case paying the bills right now, and on coding the gap is GONE.</p>
<span class="cue">[CUT]</span>
<p>This is the floor. Open weights at frontier. From here the only direction is down on price and up on capability. Stay sharp.</p>

=== THUMBNAIL_HTML_SHORT ===
<div class="thumb-spec">
  <h3>Vertical Frame (9:16, 1080×1920) — Jane</h3>
  <p><strong>Framing.</strong> Tight head-and-shoulders, eyes in the upper third, mouth at center. Face fills ~60% of frame width.</p>
  <p><strong>Expression.</strong> Quietly alarmed, slight forward lean, mouth closed. Like she just saw the numbers and is letting you in on the implication.</p>
  <p><strong>Eye direction.</strong> Direct to camera, locked. No alternate take.</p>
  <p><strong>Background.</strong> Near-black charcoal, simple. Faint teal-green gradient (#0c8d72 at 12% opacity) bleeds in from the upper-right corner. No environmental detail — the vertical viewport is too narrow for it.</p>
  <p><strong>Lighting.</strong> Key light from upper-left at ~4800K. Soft fill on the right at 25%. Subtle teal rim light from behind-right at 30% to motif the DeepSeek brand without naming it.</p>
</div>
<div class="thumb-option">
  <div class="thumb-rank">Best — Top Text Block</div>
  <div class="thumb-text">TIED CLAUDE</div>
  <p><strong>Position.</strong> Top third, single line, oversized.</p>
  <p><strong>Font.</strong> Heavy condensed sans, all caps, max display weight.</p>
  <p><strong>Color.</strong> "TIED" in #ffffff with a 6px charcoal stroke. "CLAUDE" in DeepSeek teal #0c8d72. Subtle gold underline beneath the word "CLAUDE" only.</p>
  <p><strong>Accent detail.</strong> Tiny "80.6% SWE-Bench" tag in 28pt under the headline in #c8c8c8.</p>
</div>
<div class="thumb-option">
  <div class="thumb-rank">Alternate — Stat-Led</div>
  <div class="thumb-text">$3.48 vs $25</div>
  <p><strong>Position.</strong> Top third, mono numerals stacked over a thin divider.</p>
  <p><strong>Font.</strong> Mono numerals for credibility, sans subline.</p>
  <p><strong>Color.</strong> "$3.48" in DeepSeek teal #0c8d72, "$25" in muted Anthropic-orange #d97757 with a diagonal red strike.</p>
  <p><strong>Accent detail.</strong> Small subline "per M output tokens" in 22pt #c8c8c8.</p>
</div>

=== HEYGEN_LOOK_SHORT ===
A photorealistic tight close-up portrait of a poised woman in her early 30s with a quietly alarmed expression, dark blazer, charcoal collared shirt, minimalist styling, no jewelry. Near-black charcoal background with one subtle teal-green accent gradient (#0c8d72 at 12% opacity) bleeding in from the upper-right corner, no environmental detail, vertical framing. Key light from upper-left at ~4800K, soft fill on the right at 25% intensity, subtle teal rim light from behind-right at 30%. Eyes locked direct to camera. 9:16 portrait, face fills 60% of frame width, ultrarealistic, sharp focus, shallow depth of field with subject tack-sharp and background soft, clean rendering, artifact-free, cinematic, restrained.

=== DESCRIPTION_SHORT ===
An open weights model just tied Claude on coding. One seventh the price. #DeepSeek

=== HASHTAGS_SHORT ===
#DeepSeek #SWEBench #OpenSourceAI #AINews #Shorts