Anthropic ships Claude Haiku 5.5: 75% cheaper, agent-capable mini model
智东西 · wechat · 2026-10-08
Anthropic released Claude Haiku 5.5, its cheapest and fastest mini model yet, cutting average running costs 75% vs Haiku 4.5. Pricing for requests under 100K tokens is 90% lower than Haiku 4.5 — even below DeepSeek-V4.1-Flash off-peak.
Key points:
- Beats Haiku 4.5 and GPT-6 Luna across 8 benchmarks; knowledge-work scores nearly doubled, Terminal-Bench agent coding jumped from 0% to 39.2%
- First Haiku with an adjustable effort setting to trade cost vs intelligence
- Suited for summarization, classification, real-time support, browser ops; works as a subagent alongside Sonnet 5.5 for coding
- Live on AWS, Google Cloud, Azure and Claude; Sonnet 5.5 cache reads halved to $0.1/M tokens, Max/Team subscribers get $100–$500 monthly API credits
Developer tests found it impressive on lightweight animations but up to 12x the cost of GPT-6 Luna in the same scenario, with complex multi-step planning still a weakness.
Related event: Anthropic Launches Claude Haiku 5.5 with ~75% Lower Running Costs(59 posts)→
More from Models
- 'Compression of the RH proof has begun': AI circles buzz over unverified Riemann claim — teortaxesTex · 2026-11-22
- GLM 5.3 and DeepSeek V4.1 Flash resold at 60% off: the economics of open-source giveaways — lxfater · 2026-11-22
- GPT-6.2 Astra and Sol answer more directly but score lower on transparency than Claude — The Decoder · 2026-11-22
- Baseline model test: no persona, no forced formatting, no safety bypasses observed — Sauers_ · 2026-11-22
- 2-day Gemini stress test: $4.99 Google AI Pro nearly maxed, orchestrator UI revealed — lipeng0820 · 2026-11-22
- Timed eval task asks users to deliberately make Claude fail at real work in 12 minutes — Icy-Measurement10 · 2026-11-22