DeepSeek V4 Pro Quietly Ships, Undercutting Claude Fable 5
DeepSeek quietly released the finished V4 Pro 0813 model, claiming Claude Fable 5 averages only 5.3% better across agent benchmarks. With token pricing at $0.435/$0.87 versus $10/$50, DeepSeek is roughly 46 times cheaper. Independent benchmarks are not yet available, though weights are MIT-licensed.
Quick Take
DeepSeek quietly released V4 Pro 0813 general-availability build with no announcement.
Claude Fable 5 leads by only 5.3% across nine agent benchmarks.
V4 Pro costs $0.435/$0.87 per million tokens versus Fable 5's $10/$50.
Blended rates make DeepSeek roughly 46 times cheaper; independent verification pending.
Market Impact Analysis
NeutralAI model pricing competition has no direct crypto market or digital asset implications; not a crypto-related event.
Speculation Analysis
Key Takeaways
- DeepSeek quietly shipped V4 Pro 0813 as its general-availability build, with pricing unchanged and no official announcement.
- Claude Fable 5 leads by an average of 5.3% across nine agent benchmarks, with two wins for DeepSeek.
- V4 Pro costs $0.435 per million input tokens and $0.87 per million output, versus Fable 5's $10 and $50.
- Blended rates make DeepSeek roughly 46 times cheaper, but independent benchmark verification is still pending.
What Happened
DeepSeek moved its V4 Pro model from preview to general availability on Wednesday, with no blog post or announcement. The version listed on the API pricing page now reads DeepSeek-V4-Pro-0813. Pricing remains unchanged at $0.435 per million input tokens and $0.87 per million output, but the underlying weights were updated. DeepSeek V4 Pro first appeared in April, priced 98% below GPT-5 Pro, and every independent lab that tested it was evaluating a preview. The Hugging Face model card still describes the V4 series as a preview version, so no independent benchmarks exist for 0813 yet. DeepSeek's own comparison table shows Claude Fable 5 leading by an average of 5.3% across nine agent benchmarks, with two wins for DeepSeek.
The Numbers
Claude Fable 5 costs $10 per million input tokens and $50 per million output. V4 Pro costs $0.435 and $0.87, with cached input at $0.003625. On blended rates, Fable 5 averages $30 compared to $0.65 for V4 Pro, making DeepSeek roughly 46 times cheaper. Strip out Humanity's Last Exam without tools, where Fable 5 leads by 10.6%, and the average gap across remaining benchmarks drops to 2.8%. Artificial Analysis measured Fable 5 at $3.15 per benchmark task versus $0.03 for V4-Flash, a 105x spread. Hugging Face CEO Clément Delangue put the spread at over $31 per task against roughly $0.04.
Why It Happened
DeepSeek signaled on July 31 that an official Pro release would follow soon, after moving V4-Flash to general availability. The shift from preview to GA is a routine step in the model lifecycle, but the lack of announcement suggests DeepSeek wanted to let pricing do the talking. The company scored its own benchmarks on infrastructure it has not yet released, using minimal mode settings. With weights MIT-licensed and hosted on Hugging Face, DeepSeek invites independent verification that has not yet arrived. The strategy is clear: undercut premium proprietary models on cost while claiming near-parity performance. That pressure is now public.
Broader Impact
The release intensifies pricing pressure on top AI labs. Claude Fable 5's $30 blended rate looks expensive against V4 Pro's $0.65 even with a 5.3% performance lead. Open-weight models are forcing a price correction across agentic AI. If independent benchmarks confirm DeepSeek's claims, enterprise buyers may reconsider premium subscriptions. While no direct crypto market impact exists, AI cost efficiency can influence tokenized compute and decentralized inference narratives.
What to Watch Next
- Independent benchmarks for V4 Pro 0813, using the MIT-licensed weights now hosted on Hugging Face.
- DeepSeek Harness minimal mode release, which the company said is coming soon and will enable fair comparisons.
- Anthropic's response on Claude Fable 5 pricing or Claude Opus 5 positioning, since Opus 5 outscores Fable 5 at half the price.
This article is for informational purposes only and does not constitute financial advice.
Always late to trends?
Join for the latest news, insights & more.
Disclaimer: Bytewit is an independent media outlet that delivers news, research, and data.
© 2026 Bytewit. All Rights Reserved. This article is for informational purposes only.
