← Back to live feed1 story
Anthropic's Claude Haiku 5.5 outperformed OpenAI's GPT-6 Luna on every reported benchmark, including the OSWorld 2.1 computer use test, agentic coding, and visual reasoning. The model features an expanded 1M token context window and tiered pricing that costs $0.10 per 1M input tokens and $0.50 per 1M output tokens for prompts under 100k, which is 90% cheaper than Haiku 4.5. Rates increase fivefold to $0.50 and $2.50 for requests exceeding 100k tokens, and the model uses approximately 3x more output tokens than GPT-6 Luna to reach similar intelligence index scores.
Sign in to suggest edits
Key sources
- SOURCE@claudeai“Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released”x.com
- SUPPORT@mark_k“Agentic coding: 0% → 39.2% on Terminal-Bench 4.0, versus Luna's 16.4%”x.com
- SUPPORT@rohanpaul_ai“priced 90% lower than Claude Haiku 4.5 for requests up to 100,000 tokens”x.com
- SUPPORT@artificialanlys“Context window: 1 million tokens, up from 200k for Claude 4.5 Haiku”x.com
- SUPPORT@kimmonismus“expect it close-ish to Sonnet but far cheaper, probably Luna-pricing”x.com
- SUPPORT@ns123abc“haiku 5.5 just dropped and beat it on all 6 benchmarks”x.com
- SUPPORT@theo“Charging 1/5th the price for requests under 100k tokens”x.com
- SUPPORT@teortaxestex“the deepest product lineup of all labs. Haiku/Sonnet/Opus/Fable + Mythos”x.com