MODEL VERSIONS ON THIS PAGE HAVE MOVED — CHECKED 8 OCTOBER 2026
Current line, with published API pricing per million tokens:
●
OpenAI — GPT-6 Sol (paid ChatGPT), GPT-6 Luna ($0.10 / $0.50), GPT-6.1 Sol ($2.00 / $10.00), GPT-6 Astra
●
Anthropic — Claude Opus 5.5, Sonnet 5.5 ($2.00 / $10.00), Haiku 5.5 ($0.10 / $0.50 below 100K), Fable 5.1
●
xAI — Grok 4.3 ($1.25 / $2.50), Grok 4.7 ($2.00 / $6.00)
●
Google — Gemini 3.5 Flash-Lite ($0.30 / $2.50), Gemini 3.8 Flash ($0.75 / $3.75 to 31 Dec), Gemini 4 Argon
The comparison below is kept because the reasoning still holds — how these tiers trade price against capability has not changed. The version numbers and prices have.
Current pricing table →
QUICK VERDICT — AUGUST 1, 2026
● Speed/latency: Claude Haiku 4.5 — Anthropic's fastest model
● Agentic coding: V4 Flash 0731 — Terminal-Bench 82.7%
● Price: V4 Flash — $0.14/M vs Haiku ~$0.80/M (Haiku ~6× more expensive)
● Data residency: Haiku 4.5 — Anthropic US. V4 Flash — DeepSeek (China connection on API).
● Open weights: V4 Flash only — MIT, self-hostable to eliminate data residency concern
● Use case fit: Different. Haiku = routing/classification/simple tasks. V4 Flash = agentic coding.
These two models are not in the same category. Claude Haiku 4.5 is Anthropic's fastest model — designed for latency-sensitive classification, routing, structured extraction, and simple summarisation where response time matters more than reasoning depth. DeepSeek V4 Flash 0731 is an agentic coding specialist — re-post-trained on agent data, Terminal-Bench 82.7%, designed for multi-step tool-calling coding workflows. Comparing them head-to-head makes sense only for teams choosing a budget-tier model and uncertain which type of workload it will serve.
Haiku 4.5 for: Real-time classification, chatbot routing, fast structured extraction, any task where response latency matters and quality ceiling is not the constraint. Anthropic US data residency. Zero data residency concern.
V4 Flash 0731 for: Agentic coding loops where Terminal-Bench 82.7% performance at $0.14/M is the value proposition. Self-hosted on MIT weights eliminates the DeepSeek API data residency concern. Codex-compatible by default.
Last updated August 1, 2026. Related: V4 Flash 0731 full review → · Sonnet 5 vs Haiku 4.5 →