Anthropic launches Claude Haiku 5.5: 75% cheaper, computer use score jumps from 15.7% to 72.4%
AGI Hunt · wechat · 2026-10-08
Anthropic released Claude Haiku 5.5, its cheapest, fastest and strongest small model yet, cutting average running costs 75% vs Haiku 4.5 while scoring far higher on coding, computer use and knowledge work.
Biggest gains: OSWorld 2.1 offline subset at 72.4% (vs 15.7% for Haiku 4.5, 48.9% for GPT-6 Luna), Terminal-Bench 4.0 from 0% to 39.2%, and HLE at 57.4% with tools. Pricing is tiered by prompt length: requests under 100K tokens are 90% cheaper than before, over 100K are 50% cheaper. It's the first Haiku with an effort setting (Low to Max). Anthropic also halved Sonnet 5.5 cache-read pricing (20% agent cost cut) and added monthly API credits ($100-$500) for Max and Team subscribers.
Partners like Asana, HubSpot and Cognition report lower latency and costs. The model targets summarization, context compression, subagents and latency-sensitive use cases, and is live on AWS, Google Cloud and Azure.
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