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 2026
● Input price: Both — $5/M (identical)
● Output price: GPT-5.6 Sol — $30/M. Claude Opus 5 — $25/M (Opus 17% cheaper on output)
● Terminal-Bench: GPT-5.6 Sol — 88.8% #1. Opus 5 — not published on this benchmark.
● AA Intelligence Index: Claude Opus 5 — 61. GPT-5.6 Sol — not published on AA Index.
● ARC-AGI-3: Claude Opus 5 — 30.2%. GPT-5.6 — not published.
● Output length: Claude Opus 5 — 128K tokens. GPT-5.6 Sol — 32K tokens.
● Async coding: GPT-5.6 Sol via Codex — assign and get a PR. Opus 5 — no equivalent.
● Effort dial: Claude Opus 5 only — trade cost vs capability per request.
● Safety posture: Claude Opus 5 — Anthropic FLI C+. GPT-5.6 — OpenAI FLI C.
At the same $5/M input price, the choice between GPT-5.6 Sol and Claude Opus 5 comes down to use case. GPT-5.6 Sol leads on Terminal-Bench (88.8%, the top coding agent benchmark) and has the Codex async PR delivery workflow — it is the strongest choice for teams that need an AI agent to complete a defined coding task and return a pull request. Claude Opus 5 leads on the AA Intelligence Index (61), ARC-AGI-3 reasoning (30.2%), output length (128K vs 32K tokens), and the FLI C+ safety rating — it is the stronger choice for complex analysis, long-form writing, hard reasoning, and enterprise deployments where safety posture is a procurement requirement. Opus 5 is also 17% cheaper on output ($25 vs $30/M), which matters for tasks generating large outputs.
GPT-5.6 Sol for: Agentic coding via Codex (async PR delivery), Terminal-Bench-leading performance (88.8%), OpenAI API ecosystem, DALL-E 4 image generation in the same plan. The top choice for coding-heavy workflows where Codex async execution and Terminal-Bench leadership are the deciding factors.
Claude Opus 5 for: General intelligence (AA Index 61), hard reasoning (ARC-AGI-3 30.2%), long-form output (128K), effort dial for cost control, Anthropic FLI C+ safety, enterprise compliance, and 17% cheaper output. The top choice for analysis, writing, research, and safety-sensitive enterprise deployments.
Last updated August 7, 2026. Related: Cheaper alternatives to both →