Claude Haiku 5.5 matches GPT-6 Luna pricing but a stingier tokenizer hides a 1.25x cost hike
Simon Willison · rss · 2026-10-08
- Anthropic released Claude Haiku 5.5, a fast, low-cost model priced identically to OpenAI's GPT-6 Luna ($0.10/$0.50 per million tokens) up to 100K tokens; beyond that it jumps 5x to $0.50/$2.50, while Luna only rises to $0.20/$0.75 after 272K — making Luna the better long-context deal.
- Hidden cost: Haiku 5.5 uses a less generous tokenizer — the same long prompt consumes 1.25x more tokens than Haiku 4.5, per the author's Claude Token Counter tool, effectively raising prices further.
- Hands-on testing via the llm-anthropic plugin: reasoning can no longer be disabled (defaults to medium). The pelican-on-a-bicycle test cost 0.0936 cents at low effort (7 seconds) and 3.38 cents at max effort (5m09s), with drawing quality far better than Haiku 4.5 a year ago.
- Perks: Sonnet 5.5 cache reads halved in price; Max/Team subscribers get monthly API credits ($100 for Max 5x, $200 for Max 20x, up to $500 pooled for Team), roughly equal to the subscription fee, claimable in Billing settings, with optional auto-reload disabled to avoid surprise bills. OpenAI's Codex subscription API use still wins for heavy API users.
More from coding & agent
- 3 AI agent-generated optimizations merged upstream into SGLang-Omni — ChengleiSi · 2026-10-08
- DuckDB shows a database with no data: a few hundred KB catalog over your data lake — RexDouglass · 2026-10-08
- Dev open-sources smoke effect from Replit-built Drift game with agent prompt workflow — techartist_ · 2026-10-08
- Engineer Pushes Back on AI Code Analogy: Topology-Optimized Parts Look Cool but Are Unusable — rms80 · 2026-10-08
- Massive powers AI trade agents with x402 micropayments at a cent per request — kleffew94 · 2026-10-08
- A 4-month method for using Codex App as an information-discovery engine — jxnlco · 2026-10-08