Reportedly: Anthropic releases Claude Haiku 5.5, cutting costs by 90% and undercutting DeepSeek on short outputs
Model: DeepSeek-V4.1-Flash · availability, license and releases
Reported by Zhidongxi · not yet confirmed by the company or a second independent outlet. We update this page when it is.
Anthropic released Claude Haiku 5.5, its cheapest and fastest mini model, with output pricing 90% lower than Haiku 4.5 for requests up to 100k tokens. For outputs under 100k tokens, it is cheaper than DeepSeek-V4.1-Flash idle pricing. It is available on Claude, AWS, Google Cloud, and Microsoft Azure, Zhidongxi reported.
China context
- Outside China
- Global API · zhidx.com
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Developers outside China can use claude-haiku-5-5 on Claude, AWS, Google Cloud, and Microsoft Azure for cost-sensitive tasks, and may compare its pricing directly with DeepSeek-V4.1-Flash for short outputs.
- For investors
- Anthropic's aggressive pricing on Haiku 5.5 directly targets the low-cost segment dominated by Chinese providers like DeepSeek, potentially pressuring their margins and market share in global API services.
Translated from Chinese. Quotes and facts link to the original sources.
Anthropic launched Claude Haiku 5.5, a mini model optimized for high-throughput, cost-sensitive tasks such as summarization, data extraction, database queries, classification, real-time customer support, and browser operations. It can also serve as a sub-agent alongside Sonnet 5.5 for coding. On the Artificial Analysis AI Index, Haiku 5.5 scores 43, slightly above GLM-5.3 Flash (42), Gemini 3.8 Flash (41), and GPT-6 Luna (38). Pricing is 90% lower than Haiku 4.5 for requests up to 100k tokens, with a 50% discount for larger requests. Anthropic also halved cached read pricing for Sonnet 5.5 to $0.10 per million tokens, reducing costs by about 20% in most agent tasks, and added monthly API credit bonuses for Claude Max and Team subscribers.
Claude Haiku 5.5 is positioned for high-throughput, cost-sensitive workloads and can act as a sub-agent with Sonnet 5.5. Its Artificial Analysis AI Index score of 43 places it slightly ahead of GLM-5.3 Flash, Gemini 3.8 Flash, and GPT-6 Luna in the highest performance mode. The model is available via the claude-haiku-5-5 API identifier.
Developers using short-output tasks can now get Claude Haiku 5.5 at a lower price than DeepSeek-V4.1-Flash idle pricing, directly challenging DeepSeek's cost advantage in the mini model segment. This forces DeepSeek and other low-cost providers to respond with further price cuts or feature differentiation to retain price-sensitive customers.
Claude Haiku 5.5 reduces operational costs for high-volume, repetitive tasks by up to 90% compared to Haiku 4.5, making it attractive for enterprises with large-scale summarization, classification, or customer support workloads. The lower cached read pricing for Sonnet 5.5 further cuts costs for agent-based applications.
Watch for DeepSeek's next pricing or model update in response to Haiku 5.5's undercutting on short outputs. Also monitor adoption metrics for Haiku 5.5 on cloud platforms and any independent benchmarks comparing it with DeepSeek-V4.1-Flash on real-world tasks.