Anthropic releases Claude Haiku 5.5, a small model priced 75 per cent below its predecessor
Claude Haiku 5.5 is Anthropic's fastest and cheapest model yet, aimed at summarisation, classification, subagent work and browser use. It introduces an adjustable effort setting to the Haiku tier, posts large gains over Haiku 4.5 on agentic benchmarks, and is available on AWS, Google Cloud and Azure at $0.10 per million input tokens.
Anthropic has released Claude Haiku 5.5, describing it as the cheapest, fastest and most capable small model the company has shipped. It is built for high-volume, repetitive work: summarisation, context compaction, database queries and classification, as well as speed-sensitive tasks such as live customer support and browser automation. Anthropic also positions it as a subagent that larger Opus and Sonnet models can delegate to on coding tasks.
It is the first Haiku-class model with an adjustable effort setting, letting developers trade latency against reasoning depth per request. Computer use and browser use are now supported in beta through the Python and TypeScript SDKs.
The benchmark deltas against Haiku 4.5 are large. OSWorld 2.1 rises from 15.7 to 72.4 per cent, Terminal-Bench 4.0 from zero to 39.2 per cent, and Humanity's Last Exam from 10.2 to 45.9 per cent without tools. Anthropic is explicit that Sonnet 5.5 and Opus 5.5 remain the better choices for complex agentic coding.
Pricing is $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, with higher rates beyond that, which Anthropic says works out about 75 per cent cheaper than Haiku 4.5 on average. Alongside the launch, cache-read pricing for Sonnet 5.5 was halved, and Max, Ultra and Team subscribers now receive monthly API credits. The model is available on all platforms including Amazon Bedrock, Google Cloud Vertex AI and Microsoft Azure.
Why it matters
The small-model tier is where most production tokens are actually spent, and a model at this price with genuine agentic capability makes population-scale services economically plausible. It also reinforces the orchestration pattern in which a large model plans and inexpensive models execute.
Indicops view
For Indian government workloads, cost per interaction is the number that decides whether a service reaches every citizen or stays a pilot. At these prices, a state-wide document-classification or grievance-triage service becomes a rounding error in a departmental budget. We benchmark Haiku-class models on Hindi and regional-language tasks before recommending them, since small-model performance on Indic text varies far more than English benchmarks suggest.
More in Models
Back to the newsroomGoogle launches Gemini 4 Hydrogen, a $1.50-per-million-token model for coding and agents
Google has released Gemini 4 Hydrogen, a faster and cheaper tier of its Gemini 4 family aimed at everyday coding and agentic work. It costs $1.50 per million input tokens and $7.50 per million output tokens across a 1M-token window, and is rolling out through the Gemini API, AI Studio and consumer products.
Alibaba opens a two-week preview of Qwen4-Max with a 1M-token context, but no benchmarks or weights
Alibaba Cloud has opened a time-limited preview of Qwen4-Max, its next flagship model, through the Token Plan only. It offers a 1M-token context and up to 393,216 output tokens, but the documentation lists no pricing, benchmarks or downloadable weights, and the model retires on 25 November.
Anthropic releases Claude Fable 5.2 and Mythos 5.2, 30 per cent faster and 20 per cent cheaper
Anthropic has released Claude Fable 5.2 for general use and Mythos 5.2 for vetted cyber and life-sciences organisations. The company says output is up to 30 per cent faster than the 5.1 generation, at $8 per million input tokens and $40 per million output tokens, with lower rates of safeguard blocks.