Anthropic launches Claude Haiku 5.5, price-matched to GPT-6 Luna, and halves Sonnet 5.5 cache-read price
2026-10-07
Anthropic released Claude Haiku 5.5, priced at $0.10/$0.50 per million input/output tokens under 100K tokens — matching OpenAI's GPT-6 Luna — rising to $0.50/$2.50 above that threshold. Anthropic says it costs about 75% less to run than Haiku 4.5 on average and is positioned as a cheap subagent paired with Opus 5.5 or Sonnet 5.5. The same day, Anthropic halved Sonnet 5.5's prompt-cache-read price from $0.20 to $0.10 per million tokens, added monthly API credits to Max and Team subscription plans, and built computer-use/browser-use toolsets into the Python and TypeScript SDKs. Independent testing by Artificial Analysis put Haiku 5.5 at an Intelligence Index of 43 at max effort, ahead of GLM-5.3 Flash (42), Gemini 3.8 Flash (41) and GPT-6 Luna (38), and comparable to Kimi K3 (44); Simon Willison noted Haiku 5.5 uses a less generous tokenizer, around 1.25x more tokens than Haiku 4.5 for the same prompt.
Significance 4: A flagship small-model release with a >30%-effective pricing move (price-matching a competitor plus a 75% cut from the prior generation) and independent third-party benchmark corroboration moves tracked cost/speed/reasoning axes, per the rubric's level-4 criteria, though it doesn't yet flip a verdict.
Implications · machine-drafted, not owner judgment
Anthropic matching OpenAI's cheapest-tier pricing removes the one clear cost advantage GPT-6 Luna held, and pairing Haiku 5.5 as a cheap subagent under Opus/Sonnet likely pulls more high-volume, low-margin workloads back into Anthropic's stack. Independent AA scoring puts Haiku 5.5 near Kimi K3, a tracked open-weights model, which narrows the case for defaulting to open weights purely on price for bulk work. For the owner this is a useful cost-modeling data point for AI-engineering work but doesn't change hands-on practice yet.
- An independent SWE-bench or agentic eval of Haiku 5.5 beyond Devin's single FrontierCode number
- Whether DeepSeek V4 or Haiku 5.5 wins a head-to-head on the owner's bulk-cheap cost/speed test
- Whether OpenAI responds with a further GPT-6 Luna price cut
Sources
- primary Introducing Claude Haiku 5.5 retrieved 2026-10-08
- primary Claude Platform release notes — October 7, 2026 retrieved 2026-10-08
- curator Claude Haiku 5.5 retrieved 2026-10-08
- curator [AINews] Claude Haiku 5.5 — better than GPT-6 Luna at the same pricing retrieved 2026-10-08
Models
Affected axes
Cost · Output speed · Hard reasoning