Claude Haiku 5.5 beats GPT-6 Luna at matching price, but burns ~3x the tokens
Latent Space · rss · 2026-10-08
Latent Space's AINews details Anthropic's Claude Haiku 5.5, its first Haiku update in a year:
- Pricing: matches OpenAI's GPT-6 Luna; $0.10/$0.50 per 1M tokens (under 100K), 75% cheaper to run than Haiku 4.5. Sonnet 5.5 cache reads halved to $0.10; Max/Team plans gain API credits.
- Specs: 1M-token context (up from 200K), first Haiku with effort settings and adaptive thinking.
- Independent evals (Artificial Analysis): Intelligence Index 43 (+26 vs prior Haiku), ahead of GLM-5.3 Flash, Gemini 3.8 Flash and GPT-6 Luna (38), near open-weights Kimi K3 (44); Terminal-Bench 4.0 jumps 0% → 33%.
- Caveats: 162K output tokens per task at max effort (3x Luna), 40% hallucination rate, AutomationBench only 35% due to an over-refusal bug being fixed.
- Day-one ecosystem: Cursor claims 10x cheaper short requests; GitHub Copilot and Devin (58.4% FrontierCode 1.1, 1/8 Sonnet 5 cost) shipped support; Claude SDK now bundles computer-use/browser-use tools.
- Reaction: widely read as a direct strike at GPT-6 Luna; simonw calls it far better than the 10x pricier Haiku 4.5, while some question Sonnet's middle-tier positioning.
More from coding & agent
- Dev dumps 6x-Sol over quality, burns $200 Astra sub in half a day, moves to Claude Code — Late_Change5029 · 2026-10-08
- This Solo Dev Runs His Entire Business From One Obsidian Vault With AI — dSebastien · 2026-10-08
- Cognizant to Hire 1,500 US Graduates; Devin Cut Freight Firm's Rebuild Costs 37% — shashib · 2026-10-08
- Cisco Brings Claude Managed Agents to Webex With Its Own Governance Layer — shashib · 2026-10-08
- Dev finds 6.1 Sol surprisingly good at designing native iOS apps — Dimillian · 2026-10-08
- Anthropic test of 52 devs: AI-assisted group scored 50% vs 67% without, error-finding worst — AlexTensor · 2026-10-08