DeepSeek V4-Flash is 100x cheaper to run than Claude Fable 5

Chinese AI startup DeepSeek officially released its V4-Flash model on Friday, doubling down on its signature ultra-low-cost strategy. Independent research firm Artificial Analysis reports that V4-Flash is by far the least expensive model to run among well-known AI systems — roughly 100 times cheaper than Anthropic's Claude Fable 5 on benchmark tests. The model charges $0.14 per million input tokens and $0.28 per million output tokens, with an estimated average cost of just $0.03 per test. That compares to $0.86 for Moonshot AI's Kimi K3, $1.86 for OpenAI's GPT-5.6 Sol, and $3.15 for Claude Fable 5. Artificial Analysis notes this measure is more realistic than headline pricing because it accounts for the data processed and generated to complete each task.

On performance, V4-Flash scored 50 out of 100 on Artificial Analysis' Intelligence Index, which combines nine benchmarks covering coding, reasoning, and workplace tasks. That ties Google's Gemini 3.6 Flash and sits one point behind Meta's Muse Spark 1.1 and Zhipu's GLM-5.2. However, it lags Moonshot's Kimi K3 (57) and scores nine or more points behind top-tier models like Claude Opus 5 and GPT-5.6. DeepSeek is also preparing a more powerful V4-Pro model, though no release date has been set. The launch comes as DeepSeek faces intense domestic competition from Moonshot, MiniMax, Z.AI, ByteDance, and Alibaba — the latter unveiled its largest model yet, Qwen3.8-Max, on the same day.