DeepSeek just made the AI pricing conversation a lot more uncomfortable for its Western competitors. The Hangzhou-based lab upgraded its V4 Pro model with a 1M-token context window, then promptly pointed out that Anthropic’s Claude Fable 5, which launched on June 9, scores only about 5% better on key benchmarks despite charging roughly 45 times more for API access.
The numbers that matter
DeepSeek V4 Pro is currently available in preview with API access, and general availability is expected by mid-August 2026. On the pricing front, DeepSeek is charging approximately $0.43 per million input tokens and $0.87 per million output tokens.
Compare that to Claude Fable 5’s $10 per million input tokens and $50 per million output tokens. The math is stark: for every dollar spent on Claude Fable output tokens, a developer could run the equivalent of roughly 57 queries through DeepSeek V4 Pro.
Benchmark results tell a more nuanced story, though. Claude Fable 5 leads on broader intelligence metrics, scoring roughly 59.9 compared to DeepSeek V4 Pro’s 44.3 on one widely cited index. That’s a meaningful gap on general reasoning. But DeepSeek’s model holds its own, and in some cases pulls ahead, on specific coding and analytical tasks.
