Claude Haiku 5.5 arrives with massive price cuts proving the AI pricing arms race is far from over
Claude Haiku 5.5 cuts token prices by up to 90%, but a new tokenizer takes back half the savings—and jumps OSWorld from 15.7% to 72.4%.
Why it matters
Anthropic ships a major capability jump and aggressive pricing move in Haiku, signaling the fast-model tier is now competitive enough to reshape cost-per-task economics. Practitioners evaluating Haiku for high-volume, cost-sensitive workloads need to recalculate: token unit cost dropped dramatically, but real-world token consumption per task may offset some gains due to the new tokenizer.
The key facts
5 to knowClaude Haiku 5.5 released
OSWorld computer-use benchmark: 15.7% → 72.4%
Token prices cut by up to 90%
New tokenizer increases tokens-per-task, reducing effective savings
Story key exists in 72h list: anthropic-claude-haiku-5-5-release
The story so far
Earlier coverage of this storyline
- Claude Haiku 5.5 now available on AI GatewayVercel Blog
- This story
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: Anthropic's new Claude Haiku 5.5 crushes its predecessor in benchmarks, jumping from 15.7 to 72.4 percent on the OSWorld computer use test. Token prices drop by up to 90 percent, though a new tokenizer eats into some of those savings by consuming more tokens per task.